diff --git a/.gitignore b/.gitignore index afe3b1aa6..d6bf048cb 100644 --- a/.gitignore +++ b/.gitignore @@ -64,3 +64,59 @@ target/ # Annoying stuff from contributors venv .DS_Store + +# Created by https://www.gitignore.io/api/pycharm + +### PyCharm ### +# Covers JetBrains IDEs: IntelliJ, RubyMine, PhpStorm, AppCode, PyCharm, CLion, Android Studio and Webstorm +# Reference: https://intellij-support.jetbrains.com/hc/en-us/articles/206544839 + +# User-specific stuff: +.idea/workspace.xml +.idea/tasks.xml +.idea/dictionaries +.idea/vcs.xml +.idea/jsLibraryMappings.xml + +# Sensitive or high-churn files: +.idea/dataSources.ids +.idea/dataSources.xml +.idea/dataSources.local.xml +.idea/sqlDataSources.xml +.idea/dynamic.xml +.idea/uiDesigner.xml + +# Gradle: +.idea/gradle.xml +.idea/libraries + +# Mongo Explorer plugin: +.idea/mongoSettings.xml + +## File-based project format: +*.iws + +## Plugin-specific files: + +# IntelliJ +/out/ + +# mpeltonen/sbt-idea plugin +.idea_modules/ + +# JIRA plugin +atlassian-ide-plugin.xml + +# Crashlytics plugin (for Android Studio and IntelliJ) +com_crashlytics_export_strings.xml +crashlytics.properties +crashlytics-build.properties +fabric.properties + +### PyCharm Patch ### +# Comment Reason: https://github.com/joeblau/gitignore.io/issues/186#issuecomment-215987721 + +# *.iml +# modules.xml +# .idea/misc.xml +# *.ipr diff --git a/.travis.yml b/.travis.yml index 9d066827a..4147cd431 100644 --- a/.travis.yml +++ b/.travis.yml @@ -4,8 +4,9 @@ python: - '3.5' before_install: - - apt-get install -qq python-numpy python-scipy python-matplotlib + - sudo apt-get build-dep python-scipy - pip install coveralls + - pip install scipy install: pip install -r requirements.txt diff --git a/README.md b/README.md index f7c4b737c..7b2257ec7 100644 --- a/README.md +++ b/README.md @@ -2,12 +2,20 @@ [![Build Status](https://travis-ci.org/DistrictDataLabs/yellowbrick.svg?branch=master)](https://travis-ci.org/DistrictDataLabs/yellowbrick) [![Coverage Status](https://coveralls.io/repos/github/DistrictDataLabs/yellowbrick/badge.svg?branch=master)](https://coveralls.io/github/DistrictDataLabs/yellowbrick?branch=master) +[![Code Health](https://landscape.io/github/DistrictDataLabs/yellowbrick/master/landscape.svg?style=flat)](https://landscape.io/github/DistrictDataLabs/yellowbrick/master) [![Documentation Status](https://readthedocs.org/projects/yellowbrick/badge/?version=latest)](http://yellowbrick.readthedocs.io/en/latest/?badge=latest) [![Stories in Ready](https://badge.waffle.io/DistrictDataLabs/yellowbrick.png?label=ready&title=Ready)](https://waffle.io/DistrictDataLabs/yellowbrick) A suite of visual analysis and diagnostic tools to facilitate feature selection, model selection, and parameter tuning for machine learning. + +![Follow the yellow brick road](images/yellowbrickroad.jpg) +Image by [Quatro Cinco](https://flic.kr/p/2Yj9mj), used with permission, Flickr Creative Commons. + +# What is Yellowbrick? +Yellowbrick is a suite of visual analysis and diagnostic tools to facilitate feature selection, model selection, and parameter tuning for machine learning. All visualizations are generated in Matplotlib. Custom `yellowbrick` visualization tools include: + ## Tools for feature analysis and selection - boxplots (box-and-whisker plots) - violinplots @@ -18,7 +26,6 @@ A suite of visual analysis and diagnostic tools to facilitate feature selection, - jointplots - diagonal correlation matrix - ## Tools for model evaluation ### Classification - ROC curves @@ -28,7 +35,35 @@ A suite of visual analysis and diagnostic tools to facilitate feature selection, - prediction error plots - residual plots - ## Tools for parameter tuning - validation curves - gridsearch heatmap + +## Using Yellowbrick +For information on getting started with Yellowbrick, check out our [quick start guide](https://github.com/DistrictDataLabs/yellowbrick/blob/develop/docs/setup.md). + +## Contributing to Yellowbrick + +Yellowbrick is an open source tool designed to enable more informed machine learning through visualizations. If you would like to contribute, you can do so in the following ways: + + - Add issues or bugs to the bug tracker: https://github.com/DistrictDataLabs/yellowbrick/issues + - Work on a card on the dev board: https://waffle.io/DistrictDataLabs/yellowbrick + - Create a pull request in Github: https://github.com/DistrictDataLabs/yellowbrick/pulls + +This repository is set up in a typical production/release/development cycle as described in [A Successful Git Branching Model](http://nvie.com/posts/a-successful-git-branching-model/). A typical workflow is as follows: + +1. Select a card from the [dev board](https://waffle.io/districtdatalabs/yellowbrick) - preferably one that is "ready" then move it to "in-progress". +2. Create a branch off of develop called "feature-[feature name]", work and commit into that branch. + ``` + ~$ git checkout -b feature-myfeature develop + ``` + +3. Once you are done working (and everything is tested) merge your feature into develop. + ``` + ~$ git checkout develop + ~$ git merge --no-ff feature-myfeature + ~$ git branch -d feature-myfeature + ~$ git push origin develop + ``` + +4. Repeat. Releases will be routinely pushed into master via release branches, then deployed to the server. diff --git a/docs/Makefile b/docs/Makefile new file mode 100644 index 000000000..0d0bca988 --- /dev/null +++ b/docs/Makefile @@ -0,0 +1,225 @@ +# Makefile for Sphinx documentation +# + +# You can set these variables from the command line. +SPHINXOPTS = +SPHINXBUILD = sphinx-build +PAPER = +BUILDDIR = _build + +# Internal variables. +PAPEROPT_a4 = -D latex_paper_size=a4 +PAPEROPT_letter = -D latex_paper_size=letter +ALLSPHINXOPTS = -d $(BUILDDIR)/doctrees $(PAPEROPT_$(PAPER)) $(SPHINXOPTS) . +# the i18n builder cannot share the environment and doctrees with the others +I18NSPHINXOPTS = $(PAPEROPT_$(PAPER)) $(SPHINXOPTS) . + +.PHONY: help +help: + @echo "Please use \`make ' where is one of" + @echo " html to make standalone HTML files" + @echo " dirhtml to make HTML files named index.html in directories" + @echo " singlehtml to make a single large HTML file" + @echo " pickle to make pickle files" + @echo " json to make JSON files" + @echo " htmlhelp to make HTML files and a HTML help project" + @echo " qthelp to make HTML files and a qthelp project" + @echo " applehelp to make an Apple Help Book" + @echo " devhelp to make HTML files and a Devhelp project" + @echo " epub to make an epub" + @echo " epub3 to make an epub3" + @echo " latex to make LaTeX files, you can set PAPER=a4 or PAPER=letter" + @echo " latexpdf to make LaTeX files and run them through pdflatex" + @echo " latexpdfja to make LaTeX files and run them through platex/dvipdfmx" + @echo " text to make text files" + @echo " man to make manual pages" + @echo " texinfo to make Texinfo files" + @echo " info to make Texinfo files and run them through makeinfo" + @echo " gettext to make PO message catalogs" + @echo " changes to make an overview of all changed/added/deprecated items" + @echo " xml to make Docutils-native XML files" + @echo " pseudoxml to make pseudoxml-XML files for display purposes" + @echo " linkcheck to check all external links for integrity" + @echo " doctest to run all doctests embedded in the documentation (if enabled)" + @echo " coverage to run coverage check of the documentation (if enabled)" + @echo " dummy to check syntax errors of document sources" + +.PHONY: clean +clean: + rm -rf $(BUILDDIR)/* + +.PHONY: html +html: + $(SPHINXBUILD) -b html $(ALLSPHINXOPTS) $(BUILDDIR)/html + @echo + @echo "Build finished. The HTML pages are in $(BUILDDIR)/html." + +.PHONY: dirhtml +dirhtml: + $(SPHINXBUILD) -b dirhtml $(ALLSPHINXOPTS) $(BUILDDIR)/dirhtml + @echo + @echo "Build finished. The HTML pages are in $(BUILDDIR)/dirhtml." + +.PHONY: singlehtml +singlehtml: + $(SPHINXBUILD) -b singlehtml $(ALLSPHINXOPTS) $(BUILDDIR)/singlehtml + @echo + @echo "Build finished. The HTML page is in $(BUILDDIR)/singlehtml." + +.PHONY: pickle +pickle: + $(SPHINXBUILD) -b pickle $(ALLSPHINXOPTS) $(BUILDDIR)/pickle + @echo + @echo "Build finished; now you can process the pickle files." + +.PHONY: json +json: + $(SPHINXBUILD) -b json $(ALLSPHINXOPTS) $(BUILDDIR)/json + @echo + @echo "Build finished; now you can process the JSON files." + +.PHONY: htmlhelp +htmlhelp: + $(SPHINXBUILD) -b htmlhelp $(ALLSPHINXOPTS) $(BUILDDIR)/htmlhelp + @echo + @echo "Build finished; now you can run HTML Help Workshop with the" \ + ".hhp project file in $(BUILDDIR)/htmlhelp." + +.PHONY: qthelp +qthelp: + $(SPHINXBUILD) -b qthelp $(ALLSPHINXOPTS) $(BUILDDIR)/qthelp + @echo + @echo "Build finished; now you can run "qcollectiongenerator" with the" \ + ".qhcp project file in $(BUILDDIR)/qthelp, like this:" + @echo "# qcollectiongenerator $(BUILDDIR)/qthelp/yellowbrick.qhcp" + @echo "To view the help file:" + @echo "# assistant -collectionFile $(BUILDDIR)/qthelp/yellowbrick.qhc" + +.PHONY: applehelp +applehelp: + $(SPHINXBUILD) -b applehelp $(ALLSPHINXOPTS) $(BUILDDIR)/applehelp + @echo + @echo "Build finished. The help book is in $(BUILDDIR)/applehelp." + @echo "N.B. You won't be able to view it unless you put it in" \ + "~/Library/Documentation/Help or install it in your application" \ + "bundle." + +.PHONY: devhelp +devhelp: + $(SPHINXBUILD) -b devhelp $(ALLSPHINXOPTS) $(BUILDDIR)/devhelp + @echo + @echo "Build finished." + @echo "To view the help file:" + @echo "# mkdir -p $$HOME/.local/share/devhelp/yellowbrick" + @echo "# ln -s $(BUILDDIR)/devhelp $$HOME/.local/share/devhelp/yellowbrick" + @echo "# devhelp" + +.PHONY: epub +epub: + $(SPHINXBUILD) -b epub $(ALLSPHINXOPTS) $(BUILDDIR)/epub + @echo + @echo "Build finished. The epub file is in $(BUILDDIR)/epub." + +.PHONY: epub3 +epub3: + $(SPHINXBUILD) -b epub3 $(ALLSPHINXOPTS) $(BUILDDIR)/epub3 + @echo + @echo "Build finished. The epub3 file is in $(BUILDDIR)/epub3." + +.PHONY: latex +latex: + $(SPHINXBUILD) -b latex $(ALLSPHINXOPTS) $(BUILDDIR)/latex + @echo + @echo "Build finished; the LaTeX files are in $(BUILDDIR)/latex." + @echo "Run \`make' in that directory to run these through (pdf)latex" \ + "(use \`make latexpdf' here to do that automatically)." + +.PHONY: latexpdf +latexpdf: + $(SPHINXBUILD) -b latex $(ALLSPHINXOPTS) $(BUILDDIR)/latex + @echo "Running LaTeX files through pdflatex..." + $(MAKE) -C $(BUILDDIR)/latex all-pdf + @echo "pdflatex finished; the PDF files are in $(BUILDDIR)/latex." + +.PHONY: latexpdfja +latexpdfja: + $(SPHINXBUILD) -b latex $(ALLSPHINXOPTS) $(BUILDDIR)/latex + @echo "Running LaTeX files through platex and dvipdfmx..." + $(MAKE) -C $(BUILDDIR)/latex all-pdf-ja + @echo "pdflatex finished; the PDF files are in $(BUILDDIR)/latex." + +.PHONY: text +text: + $(SPHINXBUILD) -b text $(ALLSPHINXOPTS) $(BUILDDIR)/text + @echo + @echo "Build finished. The text files are in $(BUILDDIR)/text." + +.PHONY: man +man: + $(SPHINXBUILD) -b man $(ALLSPHINXOPTS) $(BUILDDIR)/man + @echo + @echo "Build finished. The manual pages are in $(BUILDDIR)/man." + +.PHONY: texinfo +texinfo: + $(SPHINXBUILD) -b texinfo $(ALLSPHINXOPTS) $(BUILDDIR)/texinfo + @echo + @echo "Build finished. The Texinfo files are in $(BUILDDIR)/texinfo." + @echo "Run \`make' in that directory to run these through makeinfo" \ + "(use \`make info' here to do that automatically)." + +.PHONY: info +info: + $(SPHINXBUILD) -b texinfo $(ALLSPHINXOPTS) $(BUILDDIR)/texinfo + @echo "Running Texinfo files through makeinfo..." + make -C $(BUILDDIR)/texinfo info + @echo "makeinfo finished; the Info files are in $(BUILDDIR)/texinfo." + +.PHONY: gettext +gettext: + $(SPHINXBUILD) -b gettext $(I18NSPHINXOPTS) $(BUILDDIR)/locale + @echo + @echo "Build finished. The message catalogs are in $(BUILDDIR)/locale." + +.PHONY: changes +changes: + $(SPHINXBUILD) -b changes $(ALLSPHINXOPTS) $(BUILDDIR)/changes + @echo + @echo "The overview file is in $(BUILDDIR)/changes." + +.PHONY: linkcheck +linkcheck: + $(SPHINXBUILD) -b linkcheck $(ALLSPHINXOPTS) $(BUILDDIR)/linkcheck + @echo + @echo "Link check complete; look for any errors in the above output " \ + "or in $(BUILDDIR)/linkcheck/output.txt." + +.PHONY: doctest +doctest: + $(SPHINXBUILD) -b doctest $(ALLSPHINXOPTS) $(BUILDDIR)/doctest + @echo "Testing of doctests in the sources finished, look at the " \ + "results in $(BUILDDIR)/doctest/output.txt." + +.PHONY: coverage +coverage: + $(SPHINXBUILD) -b coverage $(ALLSPHINXOPTS) $(BUILDDIR)/coverage + @echo "Testing of coverage in the sources finished, look at the " \ + "results in $(BUILDDIR)/coverage/python.txt." + +.PHONY: xml +xml: + $(SPHINXBUILD) -b xml $(ALLSPHINXOPTS) $(BUILDDIR)/xml + @echo + @echo "Build finished. The XML files are in $(BUILDDIR)/xml." + +.PHONY: pseudoxml +pseudoxml: + $(SPHINXBUILD) -b pseudoxml $(ALLSPHINXOPTS) $(BUILDDIR)/pseudoxml + @echo + @echo "Build finished. The pseudo-XML files are in $(BUILDDIR)/pseudoxml." + +.PHONY: dummy +dummy: + $(SPHINXBUILD) -b dummy $(ALLSPHINXOPTS) $(BUILDDIR)/dummy + @echo + @echo "Build finished. Dummy builder generates no files." diff --git a/docs/_static/.gitkeep b/docs/_static/.gitkeep new file mode 100644 index 000000000..e69de29bb diff --git a/docs/_templates/.gitkeep b/docs/_templates/.gitkeep new file mode 100644 index 000000000..e69de29bb diff --git a/docs/api/modules.rst b/docs/api/modules.rst new file mode 100644 index 000000000..6814db03e --- /dev/null +++ b/docs/api/modules.rst @@ -0,0 +1,7 @@ +API Reference +============== + +.. toctree:: + :maxdepth: 4 + + yellowbrick diff --git a/docs/api/yellowbrick.rst b/docs/api/yellowbrick.rst new file mode 100644 index 000000000..c7b063df2 --- /dev/null +++ b/docs/api/yellowbrick.rst @@ -0,0 +1,94 @@ +yellowbrick package +=================== + +Submodules +---------- + +yellowbrick.anscombe module +--------------------------- + +.. automodule:: yellowbrick.anscombe + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.base module +----------------------- + +.. automodule:: yellowbrick.base + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.classifier module +----------------------------- + +.. automodule:: yellowbrick.classifier + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.color_utils module +------------------------------ + +.. automodule:: yellowbrick.color_utils + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.exceptions module +----------------------------- + +.. automodule:: yellowbrick.exceptions + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.regressor module +---------------------------- + +.. automodule:: yellowbrick.regressor + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.utils module +------------------------ + +.. automodule:: yellowbrick.utils + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.version module +-------------------------- + +.. automodule:: yellowbrick.version + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.yb_palettes module +------------------------------ + +.. automodule:: yellowbrick.yb_palettes + :members: + :undoc-members: + :show-inheritance: + +yellowbrick.yb_rcmod module +--------------------------- + +.. automodule:: yellowbrick.yb_rcmod + :members: + :undoc-members: + :show-inheritance: + + +Module contents +--------------- + +.. automodule:: yellowbrick + :members: + :undoc-members: + :show-inheritance: diff --git a/docs/conf.py b/docs/conf.py new file mode 100644 index 000000000..4f14dbc3f --- /dev/null +++ b/docs/conf.py @@ -0,0 +1,348 @@ +#!/usr/bin/env python3 +# -*- coding: utf-8 -*- +# +# yellowbrick documentation build configuration file, created by +# sphinx-quickstart on Tue Jul 5 19:45:43 2016. +# +# This file is execfile()d with the current directory set to its +# containing dir. +# +# Note that not all possible configuration values are present in this +# autogenerated file. +# +# All configuration values have a default; values that are commented out +# serve to show the default. + +# If extensions (or modules to document with autodoc) are in another directory, +# add these directories to sys.path here. If the directory is relative to the +# documentation root, use os.path.abspath to make it absolute, like shown here. +# +import os +import sys +sys.path.insert(0, os.path.abspath('..')) + +# -- General configuration ------------------------------------------------ + +# If your documentation needs a minimal Sphinx version, state it here. +# +# needs_sphinx = '1.0' + +# Add any Sphinx extension module names here, as strings. They can be +# extensions coming with Sphinx (named 'sphinx.ext.*') or your custom +# ones. +extensions = [ + 'sphinx.ext.autodoc', + 'sphinx.ext.intersphinx', + 'sphinx.ext.coverage', + 'sphinx.ext.mathjax', + 'sphinx.ext.viewcode', +] + +# Add any paths that contain templates here, relative to this directory. +templates_path = ['_templates'] + +# The suffix(es) of source filenames. +# You can specify multiple suffix as a list of string: +# +# source_suffix = ['.rst', '.md'] +source_suffix = '.rst' + +# The encoding of source files. +# +# source_encoding = 'utf-8-sig' + +# The master toctree document. +master_doc = 'index' + +# General information about the project. +project = 'yellowbrick' +copyright = '2016, District Data Labs' +author = 'District Data Labs' + +# The version info for the project you're documenting, acts as replacement for +# |version| and |release|, also used in various other places throughout the +# built documents. +# +# The short X.Y version. +version = '0.1' +# The full version, including alpha/beta/rc tags. +release = '0.1' + +# The language for content autogenerated by Sphinx. Refer to documentation +# for a list of supported languages. +# +# This is also used if you do content translation via gettext catalogs. +# Usually you set "language" from the command line for these cases. +language = None + +# There are two options for replacing |today|: either, you set today to some +# non-false value, then it is used: +# +# today = '' +# +# Else, today_fmt is used as the format for a strftime call. +# +# today_fmt = '%B %d, %Y' + +# List of patterns, relative to source directory, that match files and +# directories to ignore when looking for source files. +# This patterns also effect to html_static_path and html_extra_path +exclude_patterns = ['_build', 'Thumbs.db', '.DS_Store'] + +# The reST default role (used for this markup: `text`) to use for all +# documents. +# +# default_role = None + +# If true, '()' will be appended to :func: etc. cross-reference text. +# +# add_function_parentheses = True + +# If true, the current module name will be prepended to all description +# unit titles (such as .. function::). +# +# add_module_names = True + +# If true, sectionauthor and moduleauthor directives will be shown in the +# output. They are ignored by default. +# +# show_authors = False + +# The name of the Pygments (syntax highlighting) style to use. +pygments_style = 'sphinx' + +# A list of ignored prefixes for module index sorting. +# modindex_common_prefix = [] + +# If true, keep warnings as "system message" paragraphs in the built documents. +# keep_warnings = False + +# If true, `todo` and `todoList` produce output, else they produce nothing. +todo_include_todos = False + + +# -- Options for HTML output ---------------------------------------------- + +# The theme to use for HTML and HTML Help pages. See the documentation for +# a list of builtin themes. +# +html_theme = 'sphinx_rtd_theme' + +# Theme options are theme-specific and customize the look and feel of a theme +# further. For a list of options available for each theme, see the +# documentation. +# +# html_theme_options = {} + +# Add any paths that contain custom themes here, relative to this directory. +# html_theme_path = [] + +# The name for this set of Sphinx documents. +# " v documentation" by default. +# +# html_title = 'yellowbrick v0.1' + +# A shorter title for the navigation bar. Default is the same as html_title. +# +# html_short_title = None + +# The name of an image file (relative to this directory) to place at the top +# of the sidebar. +# +# html_logo = None + +# The name of an image file (relative to this directory) to use as a favicon of +# the docs. This file should be a Windows icon file (.ico) being 16x16 or 32x32 +# pixels large. +# +# html_favicon = None + +# Add any paths that contain custom static files (such as style sheets) here, +# relative to this directory. They are copied after the builtin static files, +# so a file named "default.css" will overwrite the builtin "default.css". +html_static_path = ['_static'] + +# Add any extra paths that contain custom files (such as robots.txt or +# .htaccess) here, relative to this directory. These files are copied +# directly to the root of the documentation. +# +# html_extra_path = [] + +# If not None, a 'Last updated on:' timestamp is inserted at every page +# bottom, using the given strftime format. +# The empty string is equivalent to '%b %d, %Y'. +# +# html_last_updated_fmt = None + +# If true, SmartyPants will be used to convert quotes and dashes to +# typographically correct entities. +# +# html_use_smartypants = True + +# Custom sidebar templates, maps document names to template names. +# +# html_sidebars = {} + +# Additional templates that should be rendered to pages, maps page names to +# template names. +# +# html_additional_pages = {} + +# If false, no module index is generated. +# +# html_domain_indices = True + +# If false, no index is generated. +# +# html_use_index = True + +# If true, the index is split into individual pages for each letter. +# +# html_split_index = False + +# If true, links to the reST sources are added to the pages. +# +# html_show_sourcelink = True + +# If true, "Created using Sphinx" is shown in the HTML footer. Default is True. +# +# html_show_sphinx = True + +# If true, "(C) Copyright ..." is shown in the HTML footer. Default is True. +# +# html_show_copyright = True + +# If true, an OpenSearch description file will be output, and all pages will +# contain a tag referring to it. The value of this option must be the +# base URL from which the finished HTML is served. +# +# html_use_opensearch = '' + +# This is the file name suffix for HTML files (e.g. ".xhtml"). +# html_file_suffix = None + +# Language to be used for generating the HTML full-text search index. +# Sphinx supports the following languages: +# 'da', 'de', 'en', 'es', 'fi', 'fr', 'h', 'it', 'ja' +# 'nl', 'no', 'pt', 'ro', 'r', 'sv', 'tr', 'zh' +# +# html_search_language = 'en' + +# A dictionary with options for the search language support, empty by default. +# 'ja' uses this config value. +# 'zh' user can custom change `jieba` dictionary path. +# +# html_search_options = {'type': 'default'} + +# The name of a javascript file (relative to the configuration directory) that +# implements a search results scorer. If empty, the default will be used. +# +# html_search_scorer = 'scorer.js' + +# Output file base name for HTML help builder. +htmlhelp_basename = 'yellowbrickdoc' + +# -- Options for LaTeX output --------------------------------------------- + +latex_elements = { + # The paper size ('letterpaper' or 'a4paper'). + # + # 'papersize': 'letterpaper', + + # The font size ('10pt', '11pt' or '12pt'). + # + # 'pointsize': '10pt', + + # Additional stuff for the LaTeX preamble. + # + # 'preamble': '', + + # Latex figure (float) alignment + # + # 'figure_align': 'htbp', +} + +# Grouping the document tree into LaTeX files. List of tuples +# (source start file, target name, title, +# author, documentclass [howto, manual, or own class]). +latex_documents = [ + (master_doc, 'yellowbrick.tex', 'yellowbrick Documentation', + 'District Data Labs', 'manual'), +] + +# The name of an image file (relative to this directory) to place at the top of +# the title page. +# +# latex_logo = None + +# For "manual" documents, if this is true, then toplevel headings are parts, +# not chapters. +# +# latex_use_parts = False + +# If true, show page references after internal links. +# +# latex_show_pagerefs = False + +# If true, show URL addresses after external links. +# +# latex_show_urls = False + +# Documents to append as an appendix to all manuals. +# +# latex_appendices = [] + +# If false, no module index is generated. +# +# latex_domain_indices = True + + +# -- Options for manual page output --------------------------------------- + +# One entry per manual page. List of tuples +# (source start file, name, description, authors, manual section). +man_pages = [ + (master_doc, 'yellowbrick', 'yellowbrick Documentation', + [author], 1) +] + +# If true, show URL addresses after external links. +# +# man_show_urls = False + + +# -- Options for Texinfo output ------------------------------------------- + +# Grouping the document tree into Texinfo files. List of tuples +# (source start file, target name, title, author, +# dir menu entry, description, category) +texinfo_documents = [ + (master_doc, 'yellowbrick', 'yellowbrick Documentation', + author, 'yellowbrick', 'One line description of project.', + 'Miscellaneous'), +] + +# Documents to append as an appendix to all manuals. +# +# texinfo_appendices = [] + +# If false, no module index is generated. +# +# texinfo_domain_indices = True + +# How to display URL addresses: 'footnote', 'no', or 'inline'. +# +# texinfo_show_urls = 'footnote' + +# If true, do not generate a @detailmenu in the "Top" node's menu. +# +# texinfo_no_detailmenu = False + + +# Locations of objects.inv files for intersphinx extension that auto links to external api docs. +intersphinx_mapping = {'python': ('https://docs.python.org/3', None), + 'matplotlib': ('http://matplotlib.org/', None), + 'scipy': ('http://scipy.github.io/devdocs/', None), + 'numpy': ('https://docs.scipy.org/doc/numpy-dev/', None), + 'cycler': ('http://matplotlib.org/cycler/', None), + 'seaborn': ('https://web.stanford.edu/~mwaskom/software/seaborn/', None)} diff --git a/docs/index.md b/docs/index.md deleted file mode 100644 index 6b647f28c..000000000 --- a/docs/index.md +++ /dev/null @@ -1,33 +0,0 @@ -# Welcome to Yellowbrick - -**A suite of visual analysis and diagnostic tools to facilitate feature selection, model selection, and parameter tuning for machine learning.** - -## Tools for feature analysis and selection - -- boxplots (box-and-whisker plots) -- violinplots -- histograms -- scatter plot matrices (sploms) -- radial visualizations (radviz) -- parallel coordinates -- jointplots -- diagonal correlation matrix - - -## Tools for model evaluation - -### Classification - -- ROC curves -- classification heatmaps - -### Regression - -- prediction error plots -- residual plots - - -## Tools for parameter tuning - -- validation curves -- gridsearch heatmap diff --git a/docs/index.rst b/docs/index.rst new file mode 100644 index 000000000..67aa52988 --- /dev/null +++ b/docs/index.rst @@ -0,0 +1,68 @@ +.. yellowbrick documentation master file, created by + sphinx-quickstart on Tue Jul 5 19:45:43 2016. + You can adapt this file completely to your liking, but it should at least + contain the root `toctree` directive. + +Welcome to yellowbrick's documentation! +======================================= + +**Yellowbrick is a suite of visual analysis and diagnostic tools to facilitate feature selection, model selection, and parameter tuning for machine learning.** + +Tools for feature analysis and selection +------------------------------------------ + +- boxplots (box-and-whisker plots) +- violinplots +- histograms +- scatter plot matrices (sploms) +- radial visualizations (radviz) +- parallel coordinates +- jointplots +- diagonal correlation matrix + + +Tools for model evaluation +---------------------------- + +Classification +^^^^^^^^^^^^^^^ + +- ROC curves +- classification heatmaps + +Regression +^^^^^^^^^^^^ + +- prediction error plots +- residual plots + + +Tools for parameter tuning +---------------------------- + +- validation curves +- gridsearch heatmap + + + + + + +Contents: +========== + +.. toctree:: + :maxdepth: 2 + + setup + api/modules + + + +Indices and tables +================== + +* :ref:`genindex` +* :ref:`modindex` +* :ref:`search` + diff --git a/docs/setup.rst b/docs/setup.rst new file mode 100644 index 000000000..0f8738742 --- /dev/null +++ b/docs/setup.rst @@ -0,0 +1,22 @@ +Quick Start +============== + +This quick start is intended to get you setup with Yellowbrick in development +mode (since the project is still under development). + +1. Fork and clone the repository. After clicking fork in the upper right corner +for your own copy of yellowbrick to your github account. Clone it in a directory +of your choice.:: + + $ git clone https://github.com/[YOURUSERNAME]/yellowbrick + $ cd yellowbrick + +2. Create virtualenv and create the dependencies:: + + $ virtualenv venv + $ pip install -r requirements.txt + +3. Fetch and switch to development.:: + + $ git fetch + $ git checkout develop diff --git a/examples/download.py b/examples/download.py index b878fc57d..5d11a86fb 100644 --- a/examples/download.py +++ b/examples/download.py @@ -8,7 +8,7 @@ # Copyright (C) 2016 District Data Labs # For license information, see LICENSE.txt # -# ID: download.py [] benjamin@bengfort.com $ +# ID: download.py [1f73d2b] benjamin@bengfort.com $ """ Downloads the example datasets for running the examples. diff --git a/examples/examples.ipynb b/examples/examples.ipynb index 7794e91ab..71ef47e65 100644 --- a/examples/examples.ipynb +++ b/examples/examples.ipynb @@ -6,7 +6,7 @@ "source": [ "# Yellowbrick Examples \n", "\n", - "Ths notebook is a sample of the examples that yellowbrick provids." + "Ths notebook is a sample of the examples that yellowbrick provides." ] }, { @@ -38,13 +38,21 @@ "collapsed": false }, "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/usr/local/lib/python2.7/site-packages/scipy/linalg/basic.py:884: RuntimeWarning: internal gelsd driver lwork query error, required iwork dimension not returned. This is likely the result of LAPACK bug 0038, fixed in LAPACK 3.2.2 (released July 21, 2010). Falling back to 'gelss' driver.\n", + " warnings.warn(mesg, RuntimeWarning)\n" + ] + }, { "data": { "text/plain": [ - "(,\n", - " ,\n", - " ,\n", - " )" + "(,\n", + " ,\n", + " ,\n", + " )" ] }, "execution_count": 2, @@ -53,9 +61,9 @@ }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEACAYAAACj0I2EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xl8VdW1wPHfQggEkUlBVATrs0qLjAqoqOTpC7ECIq1F\ntM4Va6s4ECccigNOKFXEh0MFFBXQqhG0moBiFJBBgSAiKtangCLOChggIev9sW8ghITce+4599xh\nfT+f+yG5uTln3WS73NnD2qKqGGOMSX31wg7AGGOMPyyhG2NMmrCEbowxacISujHGpAlL6MYYkyYs\noRtjTJqoM6GLyAQRWS8i71V5brSIrBSREhF5XkSaBhumMcaYukTTQ58E5FV7bibQUVW7AquAEX4H\nZowxJjZ1JnRVnQv8UO2511S1IvLpAqBtALEZY4yJgR9j6BcAr/pwHWOMMXGIK6GLyA1AmapO8Ske\nY4wxHtX3+o0ich5wMnBCHa+zYjEmUKoqYdzX2rYJWqxtO9oeukQe7hORk4CrgVNUdUsUQSX8MXLk\nyFDuG+a9M/E9hy1Vfo523dSKVdVb245m2eIU4G3gUBFZLSLnA+OAJsAsEVkiIuM93d0YY4xv6hxy\nUdUza3h6UgCxGGOMiUPa7hTNycnJuHtn4ntON0H9HO26qRWrV+J1rCbqG4ho0PcwmUtE0BAnRa1t\nm6B4adtp20M3xphMYwndGGPShCX0DFRUVETf/n3p278vRUVFYYdjjPGJJfQMU1RUxKAhg5jVYBaz\nGsxi0JBBgST1jRth8GBYsMD3S5sMogpXXAGvvBJ2JKnBEnqGGTNuDKU5pdAV6AqlOaWMGTfG13us\nXg3HHgtNmkD37r5e2mSYsWPhtdegd++wI0kNltCNr+bPh6OOgrPPhgkTICsr7IhMqpo+HUaPhn//\nG5o1Czua1OC5lotJTfnD8pk7ZC6llAKQXZxN/rR8X6799NPuz+NJk6B/f18uaTLU4sUwdCi8/DK0\nbx92NKnD1qFnoKKiou3DLPnD8snLq35+SWwqKuDGG2HaNJgxAw4/3I8oo2Pr0NPPmjVw9NHwwAPw\n+9+HHU14vLRtS+gmLhs3wjnnwDffwAsvQKtWib2/JfT0smGDm385+2y46qqwowlXIBuLajlTtIWI\nzBSRj0SkSERshCsDVU5+NmvmJq4SncxNeikvh9NPd3Mw+f6MAmYcr2eKXge8pqqHAbOxM0UzzoIF\n7j+8s86CiROhYcOwIzKpTBUuvxy2bYMHHwQJ5W+u1BdNtcW5IlJ9WmIg0Cfy8RNAMS7JmzRVddy9\nW8d7mTSpMxMn2uSn8cf998Nbb8HcudCgQdjRpC6vq1xaq+p6AFX9SkRa+xiTSTKVm5FK+2yGlbfx\n2qt78dD4efTvb4uDTfymT4d774W337blifHyax26zQylsTHjxlB6rMBHz8Hm49G8njz/0i1hh2XS\nwOLFcOGFUFBgyxP94LWHvl5E9lXV9SLSBvh6dy+++eabt3+ck5OTVPWDTd1KS1vB4rnwq6Vw2hB4\nvwzKwomluLiY4uLicG5eA2vb3q1ZAwMHwiOPQM+eYUcTPj/adlTLFkXkIOAlVe0U+fxu4HtVvVtE\nrgVaqGqNY+i2tCu1LVgA/ftv5qdNt1CedxeI24xUMK0g7vXrfrBli6nJlifWLZB16JEzRXOAvYH1\nwEjgReBfwIHA58BgVf2xlu+3Rp+iKnd+TpwIWVn+bkbyiyX01FNeDqecAu3awUMP2YqW2tjGIuOL\nigq46SaYMgVeeimxOz9jZQk9tajCpZfCJ5+4bf22oqV2dmKRidvGjXDaafDmm7BwYXInc5M8oq2x\nP3asa1vPPmvJPAiW0M12a9bAcce5pWOvvw6tbTGqiUK0NfZnzIB77rHqiUGyIRcDuMnPP/wBrrzS\nbbtOlXFNG3IJR9WNZt9+9y1L2yx1NfYBSqDbl93Yp9U+gJtzadUqj7w8l8xtRUt0vLRtK59rePpp\nl8ht56eJxvaNZjmuBHO99+pBmyovWA/Lli+jIq8CgLcG/x97ZS3n0UcbWTIPmCX0DFZ18vP116FT\np7AjMqlgp1OvgIr1FdQrqkcFLoHXW1bPJfOuwOa92DL7OdrtM5lBgy4KL+gMYWPoKcTPw52rT35a\nMje1qbPd7QtdOnUhtyyX3LJcunTu4p7ftgc8Nw1aLqD9wc8lNuhMpaqBPtwtTLwKCws1u3m2cirK\nqWh282wtLCz0dK3Vq1W7dFE97zzVzZt9DjTBIu0r8HZc0yMT2nZN7W7UqFG7bYuFhYXaqFm28qsH\nldaF2qjZXlpYWKiFhYWa2y9Xc/vlem67mcRL27ZJ0RTRt39fZjWYtdPEU25ZLjNfnhnTdRYudKfA\nXHGF26GXKpOftbFJ0WDV1u7yh+XvdqPZxRd/yFNPN6LHMVdw3fC/Auw07p5Mu42TlU2Kmt2aMmXH\nzk+b/DTxyMvLqzUZT58OL73UgRXvQ/v2LwLufwxVx91LKWXMuDGW0H1mCT1FxHO4s01+Gq9ibXdL\nlrjqia+8YtUTQxHrGE2sDzJgnDFRYh2DLCws1BPyBmjrfedox47f69dfJyDIBMPG0AMXbbtbvVr1\ngANUn3++5mv4NQeUKby0bRtDT1NFRUUMHHwpW+RZaFZCo5+u5MVnnkm7P3FtDD05RFM9sepmpGQq\n8JasEl6cS0RGAGcB24DlwPmqurXaa6zRh6Bn78t4Z8m1cNz9cMy9sMzbJGqys4Tuv1gTb2X1xAMP\nhIcfjn6i3RL87iW0OFfknNGhQDdV7Ywbjx/i9XrGu+rrhKdMgZJ3b4UuF0PveyHFV7KYxIm2Lksl\njRzuXF4e2+HOsd7HRCeeSdGfga3AniJSATQGvvQlKhO1nbZhq/DGwP9h7+bH8cDYFQwf8Tql+7nX\nxTKJajJX9V2gda1GGTvW2+HOsd7HRMdzD11VfwDGAKuBL4AfVfU1vwIz0dn+H8ZvG8PH/6J8z2Po\n0Ok8Lr64NwXTCrbv3rM1v8Zv06e76okvv2zVE5OF5x66iBwMXAm0B34CnhORM1V1SvXX2rmLAful\nLUycAW1KoPeZZFX0AXa/VjhV2ZmiwYp2meLixTB0qKue6GV5YjzLcNOVL2071mUxlQ9gMPDPKp+f\nDTxYw+v8XctjdnL//fMV+ULpeJUyMP7lYKm2PRtbtui7utrA7pYn+nmfTOelbXte5SIiXYCngB7A\nFmAS8I6q/m+116nXe5jdq9z5ecklS5j3jjujO57VAtXLoqbC9mxb5ZJYdrhz4oSxbPFq4DzcssWl\nwIWqWlbtNRnX6INWUQF//7urYz5jhn87P/2qF5NIltATx+vyRONNwmu5qOo9wD3xXMPEZtMmOOcc\nWL/eFdqyY+JMIlQuT9y2LbbliSaxrB56AvhVx3zNGvfnbtOmwZz5mT8sn+zibCgBSiITVcMye6Iq\nE0TTPu1w5xQR66B7rA/SdOIoWn7VsFiwQHX//VVHj1atqAgg0IhUm6jCJkXjEk37nD7dtb3PPgsp\nyAzlpW1bLZeA+TEuPWWK+3N34kQYMCCYOFOVjaHHp672uXgxnHSSq57Yo0d4cWYiq4eeJiprXKgK\n+zQdz4IF/8Xs2Vb21iTWmjUwcCA8+qgl81RhCT1gsW6g2L508FiBJZOp9/V6pjy9mk6d/jtRIZsM\nUlv73LDBHYJyxRUwaFDIQZqo2ZBLAsRSVa5v/77MKlsJ782AfZdBu7+QW9EnqZcOhsmGXOJXvX2e\neGIep5wC7drBQw/ZipawJHwdelQ3SJNGnyi9el/Gogwoe+sXS+j+UoVLL4VPPnE1WmxFS3hsDD3F\nTZkCH624l6w9zmTrns/DMqtxYRKrsnrivHmWzFORJfQkUFEBI0fCU0/BnDlZfPnlUMaM+xmA/GlW\n+N8kRmX1xLffdnsdTOqxIZeQVd35+cILtvMzVjbk4g9bnph8EnpikYnfmjVw3HGw117B7Pw0Jhq2\nPDF9xJXQRaSZiPxLRFaKyAoR6eVXYOlu4UI46ig44wyYNAkaNgw7IpMJqm/zt+WJ6SXeaouPA2+q\n6iQRqQ80VtWfq70mbf4s9cvUqW7n54QJtvMzXjbkEr3q5ZEbvdGETod+Srdurax6YhJK6LJFEWkK\nLFXV/6rjdSnV6INUdfLTz7K3mcwSevR22uavwOQH2XtDL9Z9eaStaElCiR5D/xXwrYhMEpElIvKo\niGTHcb20tmkT/PGP8MYbbrjFkrkJ1YLL4bvj6dR9lCXzNBLPssX6QHfgElV9V0TuB64DRlZ/Ybqd\nuxirtWvdwQCdO7u15jZe7p2dKerd9m3+63Jh2dU0zDqB64Y/EHZYJsKPth3PkMu+wHxVPTjy+bHA\ntao6oNrrUurP0njUtMV/4UL4/e/dmPnVV9s4pd9syCU2Dz44n+H5v6Fbjxu59aYBtschiYVxBN2b\nwFBV/VhERuImRa+t9pqUa/Re1HQe52V/mc/EiV147DHXQzf+s4QevTVr4Oij4YEHXCfDJLcwEnoX\n4DGgAfApcL6q/lTtNSnV6L3aacKpQmDaLTRaM5SFC9rQuXPY0aUvS+jRscOdU08YZ4ouA2wrQlVb\nG0PBE/BtG3r1vozOnZ8NOyKT4crL4fTT3b6HfCsLlNZsp6hP8ofl03D2IfC/c2DjRhpt68+Iq/4c\ndlgmw9nhzpnFinP5pHnzPJo0WE67Vk/S/uB/cdVlz9iEkwldZfXEuXOtemImsOJcPqjc+WmTn4ln\nY+i1mz4d/vY3Vz2xffuwozGxsnroCVZRATffDJMnw2uvYZOfJmksWQIXXgj//rcl80xiY+gebdoE\ngwfD7NmwaJG3ZF69UJIxflizxv2l+Oij0LNn2NGYRLKE7sHata7s7Z57ei97W7lufVaDWcxqMItB\nQwZZUjdx27AB+vXbUT3ROg2ZxcbQY7RokduUcdll8e383GndOkCJnR3qhY2h71Be7nrmBx4IDz8M\nM2fuutmtYFqBTdanCDvgImBTp7rez/jxcM01tgTMJI+alieOGTfGJfOuQFcozSndXprCpCebFI1C\nZdnbJ590Qyx+TH5uL5TEjt6THQZtvLLliQZsyKVOmzbBuefCunVQUODvMXE1FfMysbEhl9qXJ9ZU\nX8iGXFJHwmu5RHWDJGn0XlSWve3Uya0YqCx7a4k4eWR6Qq/rcGdrq6krlIQuIvWAd4G1qrrLtppk\naPReLFrkVglUL3trvZ7kkskJ3aonprewEvqVwBFA03RJ6FOnulUsEybsuvPTVqckl0xN6FY9Mf0l\nfJWLiLQFTsaV0E15FRVw000wYoSb/LRt/CYZWfVEU5t4V7ncB1wNNPMhllBVnfxctKj2yU9bnWLC\nZNUTze547qGLSD9gvaqWABJ5pKStW6FPH7fzc/bs3a9kycvLo2BaAbllueSW5dr4uUmoRx+FN9+E\nZ5+15YlmV/H00HsDp4jIyUA2sJeITFbVc6q/MNkP0s3Kcr2dXr2i6/Hk5eVZEg9Jph8SPWAA/O53\n0Czl/yY21YV6SPROFxHpA+Sny6SoSR2ZOilq0p9t/TfGmAxmG4tMSrMeuklX1kM3xpgMZgndGGPS\nhCV0Y9KIHWiR2WwM3aQ0G0PfweoMpRertmgyjiX0HazOUHqxSVFjjMlgltCNSRP5w/LJLs6GEqAk\nUmdomNUZyiQ25GJSmg257MwOtEgfNoZuMo4ldJOubAzdGGMymCV0Y4xJE5bQjTEmTcRzwEVbEZkt\nIitEZLmIXOZnYPEKs2Z2WPfOxPecboL6Odp1UytWr+LpoZcDw1W1I3A0cImIdPAnrPhlYnLLxPec\nblIt6aTSdVMpVq88J3RV/Spy/ByquhFYCRzgV2DGGGNi48sYuogchNtwvNCP6xljjIld3OvQRaQJ\nUAzcpqrTa/i6LdQ1gQpzHXoY9zWZI6Ebi0SkPvAy8KqqjvV8IWOMMXGLN6FPBr5V1eH+hWSMMcYL\nzwldRHoDbwHLAY08rlfVQv/CM8YYE63Aa7kYY4xJDNspaowxacISujHGpAlL6MYYkyYsoRtjTJqw\nhG6MMWnCEroxxqQJS+jGGJMmLKEbY0yasIRujDFpwhK6McakCUvoxhiTJupM6CIyQUTWi8h7VZ7r\nISKLRGRp5N8jgw3TGGNMXaLpoU8C8qo9Nxq4UVW7ASOBe/wOzBhjTGzqTOiqOhf4odrT64BmkY+b\nA1/4HJcxxpgYRVU+V0TaAy+paufI5+2Aebga6AIco6prggzUGGPM7tX3+H0TgGGq+qKInAZMBHJr\neqGdu2iCZmeKmnQVa9v2usqll6q+GLnhc0DPOoJK+GPkyJGh3DfMe2fiew5bqvwc7bqpFauqt7Yd\nbUKXyKPSKhHpAyAiJwIfe7q7McYY39Q55CIiU4AcYG8RWY1b1XIRMF5EsoDNkc+NMcaEqM6Erqpn\n1vKlXj7H4qucnJyMu3cmvud0E9TP0a6bWrF6Ffgh0SKiQd/DZC4RQUOcFLW2bYLipW3b1n9jjEkT\nltCNMSZNWEI3xpg04ak4V+T5YSKyUkSWi8hdwYVojDEmGtHsFJ0EjAMmVz4hIjnAAKCTqpaLyD7B\nhGeMMSZaXotz/RW4S1XLI6/5NoDYjDHGxMDrGPqhwPEiskBE3rB66MYYEz6vxbnqAy1U9SgR6QE8\nCxxc24tvvvnm7R/n5OQk1UJ8k1qKi4spLi4OO4ztrG0bv/jRtr2Wz30FuFtV34x8/gmuYNd3NXyv\nbb4wgbGNRSZdBbmxqHpxrheBEyI3PRRoUFMyN8aYdLZ6NaxaFXYUO0SzbHEK8DZwqIisFpHzcfXP\nDxaR5cAU4JxgwzTGmOTywgvQowfMnx92JDtYLReT0mzIxSRaaSkMHw4zZ8KUKdAroDKFVsvFGGMC\ntGKF65X/+CMsWRJcMvfKEroxxtRBFR55BHJyID/f9cybNQs7ql15XbZojDEZ4YcfYOhQ+OQTmDMH\nOnQIO6Laea7lEvlavohUiEjLYMIzxpjwzJsH3bpB27awcGFyJ3OIbshlEpBX/UkRaQvkAp/7HZQx\nxoRp2za49Vb4wx/gwQfh/vuhYcOwo6pbNEfQzY1sLKruPuBqYIbvURljTEjWroWzzoJ69WDxYjjg\ngLAjip6nSVEROQVYo6rLfY7HGGNCM2MGHHkk5ObCrFmplczBw6SoiGQD1+OGW7Y/7VtExhiTYJs3\nw9VXw0svuQ1DxxwTdkTeeFnl8l/AQcAyERGgLbBYRHqq6tc1fYMVMDJ+seJcxm8ffghDhsAhh8DS\npdCiRThxJLI410G44lydavja/wHdVbV6zfTKr9tuOhMY2ylqvFKFSZPg2mvh9tvd0kRJorEGL227\nzh56pJZLDrC3iKwGRqrqpCovUWzIxRiTQn76CS6+GN5/H4qLoWPHsCPyRzSrXM6s4+u11kE3xphk\ns3AhnHEGnHQSLFoE2dlhR+Qf2ylqjMkIFRUwejTcdx88/DAMGhR2RP6zWi5mJ0VFRfTt35e+/ftS\nVFQUdjjG+GLdOsjLg1degXfeSc9kDpbQTRVFRUUMGjKIWQ1mMavBLAYNGWRJ3aSc6p2SV16B7t2h\nd2+YPRvatQs7wuBYPXSzXd/+fZnVYBZ0jTxRArllucx8eWaoce2OrXIxVVV2SkpzSmFbFnvMvIeW\nTYfy3HPZHH982NHFJpB66DUV5xKR0SKyUkRKROR5EWnqJWBjjPHTmHFjXDJv+2t49222tWjHb7ue\nlXLJ3CuvxblmAh1VtSuwChjhd2Am8fKH5ZNdnA0lQAlkF2eTPyw/7LCMic3qs2HiPOg2EXoNIitr\nQ9gRJUydCV1V5wI/VHvuNVWtiHy6ALdb1KS4vLw8CqYVkFuWS25ZLgXTCsjL26XQpjFJacMG2Fb6\nOFIyAnqdCFnjyX4zszol0e4UbY/bKdq5hq/NAKap6pRavtfGGU1USkth69bYToKxMXQDririkCHu\nRKH+/V/jf/85GnB/daZqpySQnaJ13PAGoKy2ZF7J6l2Y3Skrg4kTYdQo+Pvf3Rbs2lgtF1NVRYVb\nV3733a5u+eDBAP/DwIH/E3ZoMUtkLZddeugich4wFDhBVbfs5nutF2NqtG0bTJ0KI0fCwQe7hB7r\nobvWQ89cX38N557rDmyeOhUOOijsiPwVyCqXymtTpV6LiJyEO9zilN0lc2NqoupKlHbuDA89BBMm\nuNrTyXaCukles2a5o+G6d4e33kq/ZO5VnT30qsW5gPXASFw99Czgu8jLFqjq32r5fuvFGMAl8pkz\n4YYbXO981Cg4+eT4KtxZDz2zlJXBjTfCU0/Bk0/CCSeEHVFwvLRt21hkEmLOHJfIv/lmx1mN9XzY\np2wJPXN8+qkrqrXPPvD449CqVdgRBSvIIRdjPFm82FW1O+ccuOACWL4c/vhHf5K5yRxTp7ohuTPP\nhJdfTv9k7pVVWzSBWLHCrVhZsMD1zC+8ELKywo7KpJqNG+Gyy2DuXDdc161b2BElN+snGV99+qnr\njZ9wAhx9NKxaBX/7myVzE7uSEndgc0UFLFliyTwaltCNL9audSfA9OzpliCuWgVXXQWNG4cdmUk1\nqvDAA5CbCzfd5MbLmzQJO6rUEM0RdBOA/sD6ynXoItICeAZoD3wGDFbVnwKM0ySpb76Bu+5y/9H9\n+c/w0Uew995hR2VS1bffurmWdetg/nx3cLOJntfiXNcBr6nqYcBsrDhXxvnxR9d76tABtmxxZzOO\nHm3J3HhXXOyGVQ47DObNs2TuhafiXMBA4InIx08Ap/ocl0lSmzbBnXfCr3/thlkWL3ZbrvfbL+zI\nTKoqL3edgzPPhMceg3vusTkXr7yucmmtqusBVPUrEWntY0wmCW3ZAo884pJ5nz5u1cFhh4UdlUl1\nn3/uEnmTJm7is02bsCNKbX4tW9zt7gorYJS6ysvd+Phtt7mt+q++Cl271vltgbHiXOnj+efhr3+F\na66B4cNtb0JoxblEZCWQo6rrRaQN8Iaq/qaW77XddCmoogKeecYVzmrbFm6/3S1DTDa2UzT1/PIL\nXHklvP662zDUo0fYESWnhBXnAmYA50U+PheYHstNTfJShRkzXC987FgYP94drJuMydyknuXLXQLf\nuNENsVgy95fX4lwvAv8CDgQ+xy1b/LGW77deTApQdT2mG290PahRo2DAgPgKZyWC9dBTgyo8/LDb\nPXzvvW7zWbK3rbBZcS7jyfz5bnv+2rWucNbgwbuOZxYVFTFm3BgguU6BsYTuP79/199/70o/fPaZ\nG2KxyfToeGrbqhrow93CJKMlS1T79VNt1071scdUy8pqfl1hYaFmN89WTkU5Fc1unq2FhYWJDbYW\nkfYVeDuu6ZGObdvv3/WcOa59XXGF6ubNPgaaAby0beuhZ6APP3R/+s6ZAyNGwF/+Ag0b1v76vv37\nMqvBLKhc3VICuWW5zHx5ZkLi3R3rofvLr9/1tm1uIn38eHeASb9+/sea7hJ+pqhJLZ99Brfc4sqP\n5ufDpEmw555hR2WSznpcYQ+A5kDL2L597Vr405+gfn038bn//j7HZ2oV18pPERkhIitE5D0ReVpE\nbH9XElq3Di69FI44wi1BXLUKrrsu+mSePyyf7OJsKAFKILs4m/xh+YHGbMLR5+g+sAQ4LPJYEnku\nStOnu3aWl+fK3VoyTyzPCT2yNn0o0E3d+vT6wBC/AjPx++47t2nj8MPdkMqHH7oNQgsXFtG3f1/6\n9u9LUVFRndfJy8ujYFoBuWW55JblUjCtIGkmRY2/3pz/JvwON+TSFfhd5Lk6bN7sOg1XXAEvvgjX\nXw977BF0tKa6eIZcfga2AnuKSAXQGPjSl6hMXH7+Ge67D8aNg9NOg2XLXM8c3AqGQUMGUZpTCsDc\nIXOjStB5eXmWxDNFjEMuK1fCkCFu9crSpdC8ecDxmVp57qGr6g/AGGA18AXwo6q+5ldgJna//OLW\n+B5yCPznP3DPPW/x6dq+XHDxjp74mHFjXDKP9MBKc0q3L1EzJpYhF1VXTOv442HYMLez2JJ5uDz3\n0EXkYOBKXE30n4DnRORMVZ3iV3AmOlu3wj//CXfcAUcdBW+8AWvX1twTN2Z3dhpyqfLcDdyw0+t+\n/BEuusgN4731FvymxsIfJtHiGXI5Epinqt8DiMgLwDHALgndChgFo7wcnnoKrrvuF8orVnDIYY9z\n0UWn0LFjHldeW6UnDpTieuL5w/KZO2QupbhEn12cTf601JngtOJcwfr2u2+hTQ3PVTF/vquQ2K8f\nTJ4MjRolMMA05kvbjnXheuUD6AIsBxrh6rw8DlxSw+sCWnafubZtU332WdXDDlM9/PDvNKvJibts\nBMntl+ueuznyOBXN7Zerqrr967n9cpNmg5BX2MYiX3Xr2U1pzPb2RGO0W89uqqpaXq56++2qrVur\nFhSEHGgG8NK2PffQVXWZiEwGFgPbgKXAo97/12LqUlhYxIib5vCfj8+jzb6tGDeuGfc+MIT3D3k9\npp64TXCa2uzTah/XVfso8kQX2KflPnz5JZx9tvur8N134cADw4zS1Caudeiqeo+qdlTVzqp6rqqW\n+RWY2dno0Yvo178pJR+dwYYO17Dm6/2AoloLHNlSQ+NF/rB8st7Lgg3ABsh6L4tje97JEUe4yc/Z\nsy2ZJzPbKZrkFi50FRDnvX0AFV2vhX5ToV4Fm/fHeuImEBVa4T7QLMq2jObBB3/L88/DcceFG5ep\nW4afEZKcioqKOPr4i2jd5m369y/ltNPgmD4XwoFPQ72KnV5rPXHjpxF/H0E55dDh17BxPqrtaHPg\n7yyZpwgrzpVkJk6cw0UXf8m2en3g13fTaM3jvPjMNICdliFmF2db8saKc/mt5f5788N+/eHDeyFn\nJDR4iJbvtOS7L74LO7SMY8W5UkzVutNnn34Db77Zh8lPdmLbIYVw6p+h4SY2l7ihlZkvz6RgWsGO\nOtXTkqcmuUkPP/8MZb9MhFW/hnNPgH3fhxJo0bxF2KGZKMWV0EWkGfAYcDhQAVygqgv9CCzdbd+C\nf9Re8PH1zDr3cIac/h9651xNceMCqKGcrY2Jm6C88w6ccQY0zCpn44YjYV0prAOKoGmnpmGHZ6IU\n7xj6WOClg2KDAAAKBklEQVQVdQdEdwFWxh9S+ikq2rUY1l3/eITS1jdC8UpoBeT9lu82/JXrhv/F\nKhuahKmogHvucZuE7roLuvd8BLqVumWLHwFdI0sZTUqIZ+t/U+A4VT0PQFXLcQW7TBXVi2HNOb2E\nwQPnMe+NSbD/s3BxV2i+xiXwsh2TnDa0YoL21Vdw7rnuwOZ33oH27WGvvSKrpnJScydxxot1J5Lu\n2CXXBVgITMKV83kUyK7hdb7voEpWNe3A3L5j84ZGSt8rlYbrtM3+r+uECW8l7bFuqQTbKepJUZHq\nfvup3nDDrkcPjho1Slvu31Jb7t9SR40aFU6AJrE7RXG9++647f7visj9wHXAyOovTLd6FzWprSxt\nRcUe8H8XwewbYb/FcEwunRrvxwUXzOSAA6wnHiur5RKfrVvdvoapU+Hpp+G//3vnrxcVFXHrnbey\ntcVWAG6981aOPPJIa5sJ4Efb9rxsUUT2Bear6sGRz48FrlXVAdVep17vkaxqOhV9l7MYl9aj42d3\n8P3Xw1j/zdtU9L4eWrxjyw19ZssWo/ef/7iJz333dccP7lPD0Hj3Xt1Z+v5S6Bt5YiZ0O7wbSxYu\nSWisJsHLFlV1vYisEZFDVfVj4ETgA6/XSxW19cS3U2DlIJh9G2vrw/SCxmzevI0x45pDWa71xE0o\npkyByy+Hm25ytctrKxnx+drPXTKvUj7380WfJyRGE79416FfBjwtIg2AT4Hz4w8pue10QAQ7imEN\nvzSfN09rxNbim0Drk1VxE9Om/IU+fToCttzQhGPjRnc03Pz57ozPbt12//oWzVvwPd/v8pxJDXEl\ndFVdBvTwKZaU9cP3h3PHHXns27I3LVrdR+s2c7nqsuGWxE2oli51R8P17g2LF0OTJnV/T9MmTWFm\nlSdmQtPDbR16qrCdojHaqRjWj92pN/cu1uzdm7vvhj/9qQn1698Udogmw6nCAw/AqFEwdqw7jCJa\ntZXPNanBarl48Oijc7lppPLTD7/hwgu/5h//+C1ZWWFHlZlsUnRn33wD55/v/p06FQ4+OLbvrz5H\nZJP44fHSti2hx+DTT+Hmm6GwEK6+Gi65BBo3DjuqzGYJfYfZs+Gcc+BPf3K98wYNvF2nplVcJvEs\noQfkiy/gttvguefcCoErr4SmNqyYFCyhQ1mZ62hMmgRPPAG5uWFHZPwQSrVFEakHvAusVdVT4r1e\nMvnmG1ffYtIkuPBC+Ogj2HvvsKMyZofPPnNj5M2aQUkJtG4ddkQmTH4ccHE5abb+/Mcf3XrdDh1g\nyxZYsQJGj7ZkbpLLs89Cz57whz/Av/9tydzEXz63LXAycDsw3JeIQrRpk1sd8I9/QP/+bqnXQQeF\nHZUxO/vlF7dJqLgYXnkFjjwy7IhMsoi3h34fcDVuf2TK2rLFJfJDDnF/ts6Z44ZZLJmbZPPee3DE\nEbB5MyxZYsnc7Cye8rn9gPWqWiIiOUCtg/fJWsCovNxNIt16K3TqBK++Cl271v19JjyZWpxLFcaP\nd5OfY8a41SwmvYRdnOsO4CygHMgG9gJeUNVzqr0uKVYCVFVRAc88AyNHwgEHwB13wNFHhx2V8SIT\nVrl8/z38+c+werVbW37ooYHf0iSB0JYtikgfIL+mVS7JlNBVYcYMN+HZuDHcfjuceGLYUZl4pHtC\nnzPHrSs/7TS4805oWMPRhCY92SHRtVCF11+HG25wY4+jRsGAAbVXnDMmbOXlrp0+8ghMmAAnnxx2\nRCYVpP3GovnzXSJfuxZuuQVOPx3q+bFY0ySFdOyhr14NZ50FWVnw5JOw336+38KkAC9tO21TW0mJ\nW3o4ZIj7k/WDD1xxf0vmJpkVFECPHq5HPnOmJXMTm7TsoX/xBRx1FFxzDVx0kY07prN06qF/8IEb\nCpwyBXr18u2yJkVZLZcqysq8FycyqSOdEjq4Mz+tcqeBBA+5iEhbEZktIitEZLmIXOb1WkGwZG5S\nkSVzE494RpTLgeGq2hE4GrhERDr4E1b8wtx8Eta9M/E9p5ugfo523dSK1SvPCV1Vv1LVksjHG4GV\nwAF+BRavTExumfie002qJZ1Uum4qxeqVL2s+ROQg3LHJC/24njHGmNjFndBFpAnwHHB5pKdujDEm\nBHGtchGR+sDLwKuqOraW1yTHvn+TtsJc5RLGfU3mSOiyRRGZDHyrqilfC90YY1JdPNUWewNvActx\n9dAVuF5VC/0LzxhjTLQC31hkjDEmMRJS2URERorIWhFZEnmcFPD9ThKRD0XkYxG5Nsh7VbvvZyKy\nTESWisiigO81QUTWi8h7VZ5rISIzReQjESkSkWYJum9Cfr+1bWZLxPuuJR5f33dQ7davdhlEmwuq\nPQXRVmq45jA/4hWRhiKyMPL7WRE5a8JbrKoa+AMYiduElIh71QM+AdoDDYASoEOC7v0p0CJB9zoW\nt1T0vSrP3Q1cE/n4WuCuBN03Ib9foA3QNfJxE+AjoEMi3nct8fj2voNst361yyDaXFDtKYi2sptr\n+hFv48i/ewALgN5eYk1k7cFErUToCaxS1c9VtQyYBgxM0L2FBP3Vo6pzgR+qPT0QeCLy8RPAqQm6\nLyTg96s1b2ZrSwLe92749b6DbLe+tMsg2lxQ7SmItlLLNSs3U8Yb7y+RDxviflc/eIk1kQn9UhEp\nEZHHAv6T+ABgTZXP15K4HawKzBKRd0RkaILuWVVrVV0PrvEBrRN470T9foGdNrMtAPZNg/cdZLsN\nsl0G1eZ8a09BtJUaNlPGFa+I1BORpcBXQLGqfuAlVt8SuojMEpH3qjyWR/4dAIwHDlbVrpGA/+HX\nfZNMb1XtDpyMq21zbMjxJGrGO6G/X9l1M1v19+nb+06Tdp3IdunHz963n2sQbaWGa8Ydr6pWqGo3\n3F8Rx4lIjpdYfTuCTlVzo3zpP4GX/LpvDb4A2lX5vG3kucCp6rrIv9+ISAHuz+i5ibh3xHoR2VdV\n14tIG+DrRNxUVb+p8mmgv19xm9meA55U1emRpwN73wls14G124Dbpe8/e7/aUxBtpaZr+tn+VfVn\nEXkFONJLrIla5dKmyqe/B94P8HbvAIeISHsRyQKGADMCvB8AItI48n9uRGRPoC/Bvk9w43ZVx+5m\nAOdFPj4XmF79G4K4b4J/vxOBD3TnncmJet878fl9B9JuA2iXQbS5oNpTEG1ll2vGG6+I7FM5TCMi\n2UAusNRTrPHMzMYwgzsZeA83c/8ibmwoyPudhJuBXgVcl6D3+KvI+1uK22wV6H2BKcCXwBZgNXA+\n0AJ4LfLeZwLNE3TfhPx+cTP/26r8nJdEftctg37ftcTj6/sOot362S6DaHNBtacg2spurhlXvECn\nyLWWAsuAqyLPxxyrbSwyxpg0YUcmG2NMmrCEbowxacISujHGpAlL6MYYkyYsoRtjTJqwhG6MMWnC\nEroxxqQJS+jGGJMm/h/y7PV8y2rhdwAAAABJRU5ErkJggg==\n", + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAFRCAYAAAA1uqfwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEbGYmq+IiIjF1HxFREQs9v8DxhmcCWUNT7sAAAAASUVORK5CYII=\n", "text/plain": [ - "" + "" ] }, "metadata": {}, @@ -128,6 +136,7 @@ "from sklearn.svm import LinearSVC\n", "from sklearn.pipeline import Pipeline\n", "from sklearn.datasets.base import Bunch\n", + "from sklearn.tree import DecisionTreeClassifier\n", "from sklearn.preprocessing import StandardScaler \n", "from sklearn.neighbors import KNeighborsClassifier\n", "from sklearn.metrics import classification_report as clsr\n", @@ -177,7 +186,7 @@ { "data": { "text/plain": [ - "" + "" ] }, "execution_count": 6, @@ -186,9 +195,9 @@ }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAERCAYAAABisfzAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n", + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdUAAAFcCAYAAACeBiTnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n", "text/plain": [ - "" + "" ] }, "metadata": {}, @@ -221,8 +230,7 @@ { "data": { "text/plain": [ - "array([,\n", - " ], dtype=object)" + "" ] }, "execution_count": 8, @@ -231,9 +239,9 @@ }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAEKCAYAAAARnO4WAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n", + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAegAAAFeCAYAAABOw4xkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3XlcVPX+P/DXrAzDsDMwgyAiinsEGrnhCmqKCigFedXS\ni1rdbLEyd03RLC0rt66lLXal3PvS4t5yrW7+yDVNMw0FZthkmYFZmJnz+wM4Ms7AsM0MyPv5ePiQ\nOedzPudzPjPwns9ZPm8OwzAMCCGEENKmcJ3dAEIIIYRYogBNCCGEtEEUoAkhhJA2iAI0IYQQ0gZR\ngCaEEELaIArQhBBCSBvEd3YDCGnLevbsifDwcHC5XHA4HGg0Gri7u2PFihXo27cvAECj0eDdd9/F\nqVOnIBQKweFwMHLkSDz11FNwcXFh6zp48CA+//xz6HQ6VFVVISoqCi+//DLc3d0t9rtt2zZ88cUX\nGDRoENauXdusti9atAjh4eF48sknzZYXFBTgueeew549e5pVb1MdOHAAu3fvhslkgtFoxIMPPoiF\nCxeCy+UiJiYGO3fuREREhNk28+bNw8CBA/HEE09ApVLhnXfewa+//goejwcOh4PHH38cU6dOdUj7\nCXEahhBSr549ezKlpaVmyz788EPmscceYxiGYQwGA/PYY48xr7/+OqPVahmGYRitVsusWbOGmTZt\nGmM0GhmGYZht27Yx06ZNY4qLi9ntVq1axUybNs3qfkePHs1kZWW1qO2vvvoqs3PnzhbV0VIXLlxg\nRo8ezZSXlzMMwzAmk4lZvnw5s2DBAoZhGGb16tXMsmXLzLZRKpVM//79GZVKxeh0OmbSpEnMhx9+\nyPZlbm4uExcXx+zbt8+xB0OIg9EImpAGMAwDps5cPkajEXl5efDy8gIAfPPNN2AYBgsXLmTLuLi4\nYMmSJUhISMCxY8cwbNgwvP/++/jyyy/h4+MDAODxeFi4cCGOHTsGg8EAPv/ur+ILL7wApVKJJUuW\nYP78+YiKisKKFSuQm5sLAEhISMDs2bORm5uLadOmISwsDLm5udi9ezf8/PxsHlNubi7i4+Nx9uxZ\nbN68Gbm5uSgoKEBeXh58fX3x9ttvQyqVIj8/H6tXr4ZCoYDBYMCECRMwZ84cAMD27dtx4sQJ6PV6\naDQavPLKK4iNjcXmzZtx9uxZFBUVoUePHhg3bhwYhkFlZSXc3d3B4XDw3HPP4fr16wCA1NRUpKSk\nYPHixRCJRACAvXv3Yvz48ZBIJDh48CDc3Nwwa9Ystv2BgYF45513UFVV1az3lJD2ggI0ITbMmDED\nHA4Hd+7cgYuLC0aOHMmedj537hwGDBhgdbtBgwYhKysLQUFBEIvFCA4ONlvv4uKC+Ph4i+3efvtt\njBo1Chs3bkTv3r0xffp0xMbGYubMmVCr1Zg2bRrkcjkiIiKgVCrx1ltvISoqqknHxOFw2J+zsrJw\n6NAhiMViPPXUU/j888/xr3/9C6+88gqefPJJjBgxAnq9HmlpaejcuTMeeOAB/PLLL/jss88gFArx\n9ddf491330VsbCwAQKFQ4KuvvgKHw4HBYMDXX3+NUaNGoUePHoiMjMSwYcMwfPhwAEBYWBh69+6N\nb7/9FgkJCWAYBgcOHMD27dsBAL///rvVY+vVq1eTjpeQ9ogCNCE2fPrpp/D09MSVK1eQlpaGyMhI\ndiQMAAaDwep2er0efD4fXC4XJpOpyftlGAYajQa//fYbdu7cCQCQSCRITEzEjz/+iIiICPD5fDz4\n4IPNO7Aa0dHREIvFAIDevXujtLQUGo0GZ86cQXl5OTZt2gSg+lr7lStXMG7cOLz++us4fPgwbt26\nhXPnzqGyspKtLyIigv0CwOfzsWHDBixcuBD/+9//8Ouvv+LVV1/FwIED8fbbbwOoHkXv3r0bCQkJ\n+P777yGXyxEeHg6g+otEc/qOkPsB3cVNiA21p7h79eqFRYsWYcmSJcjLywMAREVF4cyZM1a3OXPm\nDKKiohAWFgaDwYDbt2+bldHr9ZgzZw4KCwvr3be14MQwDHt6VygUgstt2a9x7all4O7I2mg0AgA+\n//xzHDp0CIcOHcKePXswb948XL58GSkpKaioqMDQoUORlpZmdhnAzc2N/Xnfvn04efIkpFIp4uPj\n8dprr+HAgQP49ttvUVpaCgCIi4vD7du3kZ2djb1792LatGns9g8++CDOnTtn0eYTJ07gjTfeaNFx\nE9LWUYAmpAkmTJiAqKgopKenAwDGjh0LsViM9PR06HQ6AIBWq8Xq1avh5uaG2NhYCIVCpKWlYfHi\nxSguLgZQHZzXrFkDjUYDqVRa7/7c3NwQERGBzz77DACgUqlw6NAhDB06FADMAqM1ttbXRyKRICIi\nAh9++CG733/84x84ceIEzpw5g379+uGJJ57AQw89hOPHj9c7yuXxeNiwYQP7hQYAbty4gaCgIHh6\nerJlkpOT8cknn+DKlSsYO3YsW3bMmDFQq9X48MMP2X3cunULr7/+Orp3796sYyOkvaBT3IQ0oO61\n2lpLly7F5MmTcfr0aQwZMgQ7d+7Eli1bkJSUxJ7OHjVqFHbt2gUejwcAmDNnDlxdXTF79mxwOBzo\ndDpER0dj27ZtNvf75ptv4rXXXsP+/fthMBgwadIkJCQkIDc312r76tq0aRM2b94MhmHYx79efPHF\nRh37hg0bsHr1akycOBEGgwHx8fGIj49HcXExjh49ivj4eHh7e2P8+PHIzMw0O81dKzExEVqtFnPn\nzkVVVRU4HA5CQ0PxwQcfmLX90UcfRWxsLObMmcP2GQAIBAJ89NFHeOONNzBx4kTw+XzweDw888wz\nSEhIaNRxENJecZjmfsUmhBBCiN3QKW5CCCGkDaIATQghhLRBFKAJIYSQNogCNCGEENIGOfwuboPB\niJISy7s9Sevx9hZTHzsA9bP9UR/bH/WxY0illklxbHH4CJrP59kuRFqE+tgxqJ/tj/rY/qiP2y46\nxU0IIYS0QRSgCSGEkDaoUQH6/PnzmD59usXykydPYurUqUhJScHevXtbvXGEEEJIR2XzJrEPPvgA\nhw8fNpsAH6jO4PP666/jwIEDcHFxQWpqKkaPHm2W5YcQQgghzWNzBB0SEoItW7ZYLP/rr78QEhIC\niUQCgUCA/v37W83qQwghhJCmszmCjouLQ25ursVytVoNd/e7t427ublBpVK1buuI09RO0c6wr+us\nq7Ps3onca8sxdcqBMfuvet09G9ZdZq3OutveW6f1NjIW7bHWxnvrvHdZfccNAGVcA+6Umeqvs5HH\nfe8x3m0n0+Tjvnc/9jhum3VaLDN/L+6ts6HPkHuxBuUqg2Wd9Wxnn/4y36C+/di3D5tfJ3NPZfd+\nHl1dVdBo9BZ1Oq5/Leus/7iZNti/5htU128Cw+Gy5YQ84NPkpj9m1eznoCUSCdRqNfu6oqICHh4e\njdr23ufB8tVGLDupQoXeZPXNtNp5DfzC1v1A1tfJVt+kup1u5ZfT2j7Z8myhBtrVyDob+oNlvQ3M\nPcs09Za3dRykKXTObkAHUOXsBnQAGmc3oP1iGAiryuGmVcBNo4SbVgk3jRI6F29c6ToTHAAcDiDk\nNZx1rj6NDtD3Jr0KCwtDdnY2ysvLIRKJcObMGcyePbtRdRUWmo+0j9w04KfbVeBygLrHUZuNrnYR\np84Ls2V1ylpbVrdr6i7j3LNB3TrvzeLHAcdqfbV1cnHPOs7ddY1pY93lttp97344Na9qlwkEPBgM\nRqv74dSptbH9a62N9x6btTqt9W9Lj7v2J7N+beR+Gvu5YP+38VlzdRVCWzPyaPpn7e571ti+qK3T\n7sdd80PDv2McK8uauJ9G/C57uIugUmnNP2uN/Iw23IfmvzOO/uw0pc6Gj9v86Jrz2fH2FqO0pNKJ\nx13nt7vRx33PsdXTxto67fU7o1DkYuvWTWb7E4lcEdY5EJuTXdFSjQ7QtR+EzMxMaDQaJCcnY9Gi\nRZg1axYYhkFycjL8/f2b1YgyXXXwXz1EiMGB9NB8S0ml7hZfgkjro362P6nUFYWFBmc3474m9ROg\nkKEnbutSqcqhVCqgVOZBociDWq3CrFnzLMr5+8vQt+8DCAgIhFwuh0wWCE9PL5t52hurUQG6U6dO\nyMjIAADEx8ezy0eMGIERI0a0uBFlNWcKPYUtrooQQghpFqPRiA0b0qFSlZstFwpdoNVqIRKJzJbz\n+XykpMywW3scPhe3NWX66hG0p0vrfOsghBBC6qqsrIRSmceOisePnwRXV7FZGR6PB3//AAQFBUMm\nC6z5J4e3tw+4XMefZWgbAVpHAZoQQkjr+/LL/fjjj8soLy8zWx4V9RBCQ8Msyj/55FxHNc2mNhKg\nAS4HcBM4uyWEEELaC61Wi/z86mvFYWHh8POTWpRRq9VgGAbh4T3ZEbFMFghfXz8ntLhp2kSALtcz\n8BQC3Fa6sE4IIeT+dOHCWVy8eB5KZR5KSu6wyydOTLIaoB99dBr4/DYR6pqsTbS6TMfAW0TBmRBC\nOjq9Xo+CAiVcXESQSi2fDCooyMeVK5cgFruha9fu7N3T1k5XA2i3wRloAwHaaGKg0gNdPClAE0JI\nR1NYWIDLly9CqVRAochFcXERGIbBww8PxsSJSRblBw4cgujoQXB392i1x5naKqcHaJW+egYresSK\nEELuTwaDARpNJdzdLWebVChycezYNwAAkUiEzp27QC4PRPfuPazWJZE0fcrM9srpAZoesSKEkPuH\nXq/DrVvZ7ONMSqUChYX5CAkJxezZT1mU79q1G6ZNewIyWSC8vLzv+1FxU9gM0AzDYOXKlbh69SqE\nQiHS09MRHBzMrj927Bi2b98OLpeLpKQkpKamNqkB9IgVIYS0PwzDWA2md+7cwUcf/Zt9LRAI0KlT\nMIKCgi3KAtUj4l69+tqtne2ZzQB9/Phx6PV6ZGRk4Pz581i3bh22bt3Krl+3bh0OHz4MkUiECRMm\nID4+3izLlS13ZxGjAE0IIW2RRlPJjoZrJ/uorKzESy8tsSgrlfpj+PDRkMurJ/rw8fF1yiQf9wOb\nATorKwsxMTEAgIiICFy6dMlsvUAgQFlZGftNqqmnJ+6OoJu0GSGEEAcwGAxYt24lTCYTu4zP58Pf\nPwA6nQ4uLuZ/vHk8HuLiHnF0M+9LNgP0vXmf+Xw+TCYT+41o1qxZmDJlCsRiMeLi4iCRSJrUALoG\nTQghjqfT6ZCfr8Dly3fw5583oFQq8I9/zIKbm5tZOT6fj6ioaLi6itjpL/38pODxKLGRvdkM0BKJ\nBBUVFezrusFZoVBg9+7dOHnyJMRiMV566SUcOXIEY8eObXQD2FPcFKAJIcQhPv54B65fv2aWRpjL\n5aK4uNAiQANAQsJURzaP1LAZoKOionDq1CmMGzcO586dQ3h4OLtOp9OBx+NBKBSCw+HAx8cH5eXl\nDdRWTSq9OyLXccoBGBAql0DqQd/IWkvdPib2Q/1sf9THjafX65GXl4ecnBzk5ORgyJAhZjf11pLL\nA8DlAkFBQew/uVwOgYDmW25LbAbouLg4nD59GikpKQCqbwqrmxM6ISEBKSkpNc+vdUZiYqLNndbN\noZtfXp3w3qiuQKGORtGtgfIUOwb1s/1RHzfOTz/9gDNnfkFRUaHZqFgs9oRI5GVRfsyYSezPtX1c\nWqoFoHVEczuk5nzRtBmgORwOVq1aZbYsNDSU/fmJJ57AE0880eQd1yrTMRBwAVenP5FNCCFtj8Fg\nQFFRARQKBXx9fdG5cxeLMlqtFuXlZQgODmHvnpbLA+HvL3N8g0mrcXpYLNdV38FND6cTQki1W7f+\nxq+//sxO8mE0GgEADz000GqAjokZiREjYulxpvuM0wN0mZ5BgJiCMyGk4zCZTCgqKkRVVRU6dQqy\nWF9eXoZz57IgEAjYEbFMFoiQkC5W66Nrx/cnpwboKiODiiq6g5sQcn9Tq1W4dOl8zWQfecjPV8Jg\nMCAkJBRpac9YlO/WrQeee+4V+Pr60ai4A3NqgK65P4wCNCGk3TOZTKioUFtNCFFZWYnMzEMAqify\n8PcPgEwWiODgEKt1iUQiiEQiu7aXtH1ODdDsLGKUyYoQ0o6YTCbk5t5mR8TVU2AqwOPxsHjxKot7\navz8pJg6NRUyWSCkUn+a5IM0inMDNM0iRghpw+pLCGEymfDBB1vZm7e4XC78/KSQyQJRVVUFodB8\n1MHlcvHgg/0d0mZy/3DyCLr6fwrQhBBnq6qqQmFhvkVSiPnzX7Y4bc3n8zF69Fi4uUkglwdCKg2g\nG7VIq2sbp7gpUQYhxMnef/89KJV57Ovq2RF9oVarrF5XHjZslCObRzqgthGgKdUkIcQOjEYjO8lH\n7Yh45Mg4hISEWpTt06cfOncOMZvk495MTYQ4ks0AzTAMVq5ciatXr0IoFCI9Pd1sbtcLFy5g/fr1\nAICAgACsX7++0ad66Bo0IcRejhz5Cj/99AN7nbhWnz4PWA3QI0fGOapphDSKzQB9/Phx6PV6ZGRk\n4Pz581i3bh22bt3Krl++fDnee+89BAcHY+/evcjJyTGbCrQhd69BN6/xhJCOx2Qyobi4iL17Ojg4\nBD179rYoJ5G4QyaTs5N81P7s6urqhFYT0nQ2A3RWVhZiYmIAABEREbh06RK77ubNm/Dy8sKuXbvw\n559/YsSIEY0OzsDdU9wedIqbEGLDH39cxnffHUd+vgJVVVXs8v79o60G6CFDhmHIkGGObCIhrcpm\ngFar1XB3v5uFg8/nszmhS0pKcO7cOaxYsQLBwcGYO3cu+vbti4cffrhROy/TMxDxABGfAjQhHRnD\nMCgpuQOlMg9cLs9qwGUYExSKXEilATWjYXnN9eJOTmgxIfZnM0BLJBJUVFSwr2uDMwB4eXmhc+fO\n7Kg5JiYGly5dshmga9Nuqat08HblUL5XO6A+dQzq5+YrKirCsWPH2NzFWm11qsPQ0FDExNz9G1Lb\nx97eD2Hw4IfA5zs9hcB9hz7HbZPNT3pUVBROnTqFcePG4dy5cwgPD2fXBQcHo7KyErdv30ZwcDCy\nsrIwdepUmzutze9aojEh2IND+V5bGeXQdQzq54YxDIPy8jKUlZVazcBUXKzCd999Bw6HAz8/Kbp3\n7wmZLBCdOgWx/Up9bH/Ux45hl3zQcXFxOH36NFJSUgAA69atQ2ZmJjQaDZKTk5Geno4XX3wRABAZ\nGYnhw4c3asdaAwOtkR6xIuR+YTAYcOHCWSiVeVAo8pCfr0BlZSVcXERYunS1xYxc3t4+mDdvPvz9\nZRYzbxFCGhGgORwOVq1aZbas7o1gDz/8MPbu3dvkHZfTI1aEtEtqtQpubhKLgMvlcvF//3eAvYHL\nx8cXXbqEQS4PhMFgsHj8ksvlIiios8PaTUh747SLOfSIFSFtX0FBPhSKXPaRJoUiD2q1Ci+9tBRe\nXl5mZblcLqZMSYGHhycCAmRwcaFsTIS0hBMDNM0iRkhbt3fvZ1Ao7k5/6enphR49esFgqLJavm/f\nCEc1jZD7nvMCNJ3iJsThTCYT7twprpMisfp6cUJCMrp372FRfuDAodDrdexEH66uYie0mpCOqQ2c\n4qYATYijHDz4Bc6e/X9myyQSd2i1Gqvl+/ePdkSzCCFWOP8UN12DJqRFGIZBaWmJWYrEHj16Iyrq\nIYuyXbt2g8lkglx+d/pLiYSegSWkLXJ+gKZr0IQ029mzWfjqq0MWI2Cx2M1qgI6MHIDIyAGOah4h\npAWceA26+n86xU2IJYZhoFKV11wjVkAsFuOhhwZalHNzc4NEIkG3buE1015WT3/p4eHphFYTQlqT\n00fQHjQ/ASGs/HwlvvrqMJTKPFRW3p1iNzAwyGqADg/vifDwno5sIiHEQVqcD7rW8uXL4eXlxc4q\nZkuZjoGbABDwaARNOo6KigooldXPEkdERFmsFwqFuHHjT3h7+yAkJBQymZy9XkwI6VhanA8aADIy\nMnDt2jVERzf+js9yPUPXn8l9T6/X47vvjrOPM6lU5QCqA3G/fg+yiWdqeXl5Y+nS1RCJKGcxIR1d\ni/JBA8DZs2dx8eJFpKSk4MaNG43aKcMwKNMBYV62yxLS1mm1GiiVCoSEhFpMf8nn8/HLL6eh1+vg\n4eGJHj16sXdPMwxjUReHw6HgTAgB0MJ80IWFhdi8eTO2bt2Kr7/+utE71RiAKhPdIEbapz/+uIzc\n3Nu4c6cA2dm3UFpaAgB48cVF8PHxNSvL5XIxe/Y8eHv7QCx2c0ZzCSHtVIvyQX/77bcoLS1FWloa\nCgsLodPp0LVrVyQkJDRYJ0/iBkCLAA8h5SG1E+rXltHpdOByuRYJHgBgx44TyM7OBgC4u7ujV69e\nCAoKglTqAR8fy36XSnvbvb33M/os2x/1cdvUonzQ06dPx/Tp0wEABw8exM2bN20GZwC4qVADAFwY\nA+UhtQPK79o05eVlyMvLqTPRhwLFxUWYNu1J9OxpGVyHD48Dh8NBnz7dodffvYZsNIL6vZXRZ9n+\nqI8dwyn5oJuDZhEjbcmxY9+YTX8pErkiJCQUPB7Pavnax5o8PekPGyHEflqcD7pWYmJio3dKs4gR\ne1OrVVAo8ti7p5XKPPTvH40hQ4ZblO3b9wH4+Pixk3x4enpZ3OxFCCGO5pSJSihRBrGnn3/+EV99\nddhsmVAohFartVq+R4/e6NGDrhMTQtoW5wRoPZ3iJk1XWVnJJoNQKPLg5eWN0aPHWpSTy4PQs2cf\ns0k+vL19LJ45JoSQtsxJI2jKBU0aLyfnFv7zn49RXl5mtrxTp2CrAbpLl1B06WJ5GYYQQtoT5wZo\nugbd4Wm1WiiVCuTn50Gj0WDEiFiLMu7uHmAYBt2794Bc3gkyWfW1Yl9fPye0mBBCHMNJp7gBDgB3\nSpTRIel0WuzblwGlMg8lJXfY5Xw+HzExIy3unvb09MLChcsd3UxCCHEqp42g3YUAj0sj6PuRXq9H\nQYESSqUCUVEPWVz7FQpdcPPmdXC5PHTt2p29e1omC6S7pwkhpIZTAnS5jqHrz/eZn376AbduZUOp\nzENxcRE7z3SXLl3h5yc1K8vhcPDii4vg6iqmgEwIIfVweIA2MQzK9ECghP4wtycGgwGFhfnw9vax\nmszh4sXzuH07GyKRiE2TKJMFQiwWW62P5qUmhJCGOTxAq3QMTAw9YtXW5eTcwt9/36iZ5EOBwsJ8\nmEwmPP74E+jdu69F+cmTp8DFRQQvL28aFRNCSCuwGaAZhsHKlStx9epVCIVCpKenIzg4mF2fmZmJ\nTz75BHw+H+Hh4Vi5cmWD9ZVqTQDoEau2wGg0wmg0QCi0/LZ05swvyMr6FUD1JB+dOgVDJpPDw8PD\nal0yWaBd20oIIR2NzQB9/Phx6PV6ZGRk4Pz581i3bh22bt0KoDrjz7vvvovMzEwIhUIsWLAAp06d\nwsiRI+utr0RLj1g5g1arhUKRWzPJR3VSiIICJUaMiMOIEaMtyg8Y8DC6d+8BmSwQPj6+NMkHIYQ4\nmM0AnZWVhZiYGABAREQELl26xK4TCoXIyMiAUFj9vJTBYICLS8Pnrks1tSPoZreZNENW1q/45psv\n2dd8Ph/+/gEQiy2vJwNAcHAIgoNDHNU8QkgH9dlnH+OLL/Zg377/g0AgwNq1qxAbOxbR0QPZMpMn\nj8Xhw0cAAD/88B327csAwzDQ6/VITf2H1UGGLV9+eRBffnkQfD4fM2bMwuDBQ83WX7/+JzZsWAce\nj4egoGC8/PJi8Pl8ZGTsxtGj38DFxQVJSY8iLm5cyzqgATYDtFqthrv73TRZfD6fzQnN4XDg4+MD\nAPj000+h0WgwePDgBusroVPcrUan0yI/X8kmg1AqFQgIkCEtbZZF2a5dwxATM4J9nMnPT1pvtiZC\nCHGUo0e/RWzsWBw/fgSPPBJfT6nqeHHx4nns3bsHGza8AxcXEcrLyzB37pMIDQ1DSEiXRu/zzp1i\n7N//OT78cDd0Oi2efvqfiI4eCD7/bkhcv341nn/+FfTp0xc7dmzDgQN7MWBANI4c+Ro7dnwCk8mE\n2bP/gQEDouHt7dOCHqifzQAtkUhQUVHBvq4NzrUYhsEbb7yB7OxsbN682eYOS2pG0J2lYkilNIxu\nrmvXrmHjxo1my7hcLnx9vQFY5h6VSnvigQd6Oqx9HQUlurc/6mP7c1Yf//rrrwgLC8WsWTPw8ssv\nY8aMVIhEAnh6upq1icvlQCp1x7FjX+Gf/5yFoCAp2+6DBw+YDSIBYOnSpbh16xb72svLC++++y77\n+uLFM4iOfghyefXfy7CwriguzkXfvndvgC0qKsSIEYMAAEOHDsTnn3+OsLDOGDx4ELtdr149cfv2\ndYSHN30E3xg2A3RUVBROnTqFcePG4dy5cwgPDzdbv2zZMohEIva6tC2lNdegodWgsFDf9Bbf56qq\nqlBQoGTvntbr9UhKetSinEAgQWhoGJsMQiYLhL9/APsNkPIU2x8lurc/6mP7+/gaB99c07RqnSOC\neZgXIbBZbvfu/yAubgLc3HwBcPH99z9Dq61CWZnG7H1nmOq/aTk5CkgkvhafCa3W/PVzzy202Ffd\nbRSKIvB4LuwyLleAnJwCBATcLSOTBeL48R8QERGJr78+ivJyNaTSTvj55224dasAer0O/+//ZeGh\nhwY36jPanC9BNgN0XFwcTp8+jZSUFADAunXrkJmZCY1Ggz59+uDAgQPo378/pk+fDg6HgxkzZiA2\n1nI+5Vp0its6jUaDf//7PRQVFbKTfADVlxQmT55icTpaInHH7NlPObqZhBDSKlQqFX7++SeUlJRi\n377PUVFRgf37v4BY7Aa93nzwZjQaAQAymQwFBUqEhXVj1128eB4+Pr7o1CmIXbZ+/Rrk5NxmX3t6\nemLNmjfY12Kxm9mZ4crKSkgk5gF00aLleOedjTAad+CBBx6EWq1C585dkJSUjAULnkVAgAx9+vSD\nl5dX63SIFTYDNIfDwapVq8yWhYbezRR0+fLlJu3w7k1iHSdAGwwGFBUVQKGoTgoRFzfeIuCKRCIY\nDAZ07tyFneRDLg+Ev7+MrhUTQuzmpSESzAxnbBdsZUeOfIX4+Ml4+un5AKrvqUlOnozU1On4/vuT\nGDp0GACE5ZXYAAAgAElEQVTg/PmzbHa68eMn4f33NyMycgBEIhFKSu5g7dpVSE9/w6zuhQuXNrjv\n3r37YMeObaiqqoJOp8OtW3+ja9cwszI//fQjVqxYAw8PD2za9CaioweitLQUlZUV2Lr1A1RUqDFv\n3iz06dOvtbrEgsMnKinRMuByADfbZz/avczMQ/j77xsoLMxnvwECQFRUNPz9A8zK1k5/SZN8EEI6\ngq+++hLLlr3GvnZxEWHEiFHQ6bQQi8V48snH4eYmAZ/PxyuvLAEA9O3bD5MmJeKFF54Gj8eHXq/H\nU0/NR9eu3erbjVU+Pr5ITn4MTz89GwwDzJnzDAQCAf7++yYOHPgCL764EMHBIXj++acgFLqgZ8/e\nGDduAjgcDm7dykZa2gxwuTw8/fR8u86KyGHqnk91gPjPilGmMeLAZOuP97QXJpMJxcVFUCrzEBoa\nZnF6BAB27NiCvLwcBATI2OvEMlkgOnUKgkBgv28odN3OMaif7Y/62P6ojx3DLtegW1uJxgSvdnp6\n+/ffL+DatT+gVOYhP18Jg8EAAEhJmY6+fSMsyj/++BNwdXWlST4IIYQ0mcMDdLmOQYhH2wxYJpMJ\nJSV3IBAI4OHhabH++vVryMr6FTweD/7+AeyIWC7vZLU+NzdKCEEIIaR5HB6gGQCeQkfv1bri4iLc\nuHG9ZgpMRc1jTTqMGjUGo0aNsSg/ZMhwPPzwEEil/nTjFiGEELtySj5oR97BzTAMqqr0VhNCXLv2\nB7766hCA6kk+/Pyk7KjYmnvzGhNCCCH2cl8FaIPBYDbJh0KRi/x8BcLCwpGSMt2ifPfuPZCY+Cg7\nyYc9b9wihBBCmsI5AdpOmayys29i16732dfVc4X7Wr2eDFSPiGlUTAghpC1qcT7okydPYuvWreDz\n+ZgyZQqSk5Nt7rQpmayMRiMKCwtqkkFUj4yNRqPVWbTk8kBERw8ym+TDVnYtQgghpC1qUT5og8GA\n119/HQcOHICLiwtSU1MxevRoNsNVfRp7iruyshLr168ym+QDAHx9/SySdgDV07dNmjSlUXUTQggh\nbVmL8kH/9ddfCAkJgUQiAQD0798fZ86cwdixYxus013A1BkVK5Cfr0Rq6gyLO6PFYjFCQ8Pg6elV\nc/NW9RSYrq7te5ITQgghxJYW5YO+d52bmxtUqoZnpOl18xN8seVvGKqqzJYXFRUiIEBmUf6JJ+bY\nPAhCCCHkftOifNASiQRqtZpdV1FRAQ8Pjwbr27vhpea2lTQB5dB1DOpn+6M+tj/q47bJ5pReUVFR\n+P777wHAIh90WFgYsrOzUV5eDr1ejzNnzuDBBx+0X2sJIYSQDsJmsoy6d3ED1fmgf//9d2g0GiQn\nJ+O7777D5s2bwTAMpk6ditTUVIc0nBBCCLmfOTybFSGEEEJsa5tZKwghhJAOjgI0IYQQ0gZRgCbt\nTs+ePTFp0iQkJCQgMTER48aNQ3Jystkz+hqNBuvXr8e4ceMwadIkTJ48GZs2bYJOpzOr6+DBg0hJ\nSUFiYiLi4+OxfPnyeh8V3LZtG0aOHInFixc3u+2LFi3Crl272NcqlQr/+Mc/8Nxzz0Gv1+PVV1/F\nI488Aq1Wa7ZdZGQk8vLyGqz75MmTSE9Pb7BMbm4uIiMjra7bvHkzVq5c2bgDaQaTyYRdu3ZhypQp\nbH9v2LABer0egGXftIa6ffLHH38gLi4OSUlJ2L17t82+IsTZnDIXNyEtweFw8Omnn8LT8+4c6zt3\n7sSaNWuQkZEBo9GIJ598EpGRkTh8+DBcXFyg0+mwYcMGzJ49G5988gm4XC62b9+O//73v9i6dSt8\nfHxgNBqRnp6Op556Crt377bY7/79+7Fx40ZERUW1ynEUFxcjLS0NUVFRWLp0KXtseXl5SE9Px+rV\nq82O2ZZRo0Zh1KhRNss1pi57WLFiBVQqFT7++GNIJBJotVosWLAAy5Ytw/r16+2yz7p9cuLECQwc\nONCsXwlpyyhAk3aHYRjUvbfRaDQiLy8PXl5eAIBvvvkGDMNg4cKFbBkXFxcsWbIECQkJOHbsGIYN\nG4b3338fX375JTs1LY/Hw8KFC3Hs2DEYDAbw+Xd/PV544QUolUosWbIE8+fPR1RUFFasWIHc3FwA\nQEJCAmbPno3c3FxMmzYNYWFhyM3Nxe7du+Hn52dxDAqFArNmzUJiYiLmzDGfjGfGjBk4dOgQjh49\nijFjxrDHXOu3337Dxo0bodFowOVy8eyzz2L48OE4ePAgjhw5gu3btyM7OxtLlixBWVkZpFIpGIbB\n5MmT8dBDD8FoNGLFihW4ePEiVCoVXnnlFcTFxQEAbty4genTp6O0tBS9e/fGihUrIBaL8eeff2L1\n6tUoLS0Fl8vFE088gYSEBPz6669IT0+Hq6srtFotdu/ejSVLluDWrVvgcDjo27cvXnvtNdy+fRuZ\nmZk4ffo0xGIxAEAkEuG1117D2bNnLfpn3759+OKLL2AwGFBaWoq0tDSkpqaiqKgICxcuRElJCQBg\n+PDheO655yyWjxgxAvPnz2f7ZMKECdizZw9MJhO0Wi0GDx7M9pVarUZ6ejquXbsGg8GAQYMG4ZVX\nXgGXy0W/fv0wevRoXL16FRs2bECfPn0a9RklpDVQgCbt0owZM8DhcHDnzh24uLhg5MiRWLt2LYDq\n5/UHDBhgdbtBgwYhKysLQUFBEIvFZolfgOpAHh8fb7Hd22+/jVGjRmHjxo3o3bs3pk+fjtjYWMyc\nORNqtRrTpk2DXC5HREQElEol3nrrrXpH2jdu3MDjjz8OHo+HmTNnWqz38fHB+vXr8cILLyAiIgIB\nAQHsuvLycixevBg7d+5EYGAgCgoK8OijjyIjI8OsjoULFyIxMRGPPfYY/vrrL0ydOhWTJ08GAOh0\nOsTExGDVqlU4fvw41q9fzwbonJwcHDhwAF5eXnj55Zexbds2PP/883j66aexcOFCxMbGoqCgAMnJ\nyQgNDQUAXL9+HSdOnIBMJsPhw4dRWVmJgwcPwmQyYeXKlbh9+zauXLmC7t27s8G5lq+vL2JjY82W\nVVZWYt++fdixYwc8PT1x/vx5PPnkk0hNTcUXX3yB4OBgfPjhh9BoNFi6dCnUanW9y2tNnDgR2dnZ\nKC0txdKlS3Hw4EF23dq1a9G3b1+sW7cOJpMJr776Knbt2oXZs2ejqqoKo0ePxqZNm6y+l4TYEwVo\n0i7VnuK+cuUK0tLSEBkZaZakxWAwWN1Or9eDz+eDy+XCZDI1eb8Mw0Cj0eC3337Dzp07AVTPqJeY\nmIgff/wRERER4PP5DU7Yk5mZibfffhufffYZFi9ejI0bN1qUGTx4MJKSkvDSSy/hk08+YZefPXsW\nhYWFeOaZZ9hRNZfLZecpAKqD+IULF/DZZ58BqJ5QaODAgex6oVDIBsWePXvizp077LoxY8awZyKS\nkpLw5ptvIiEhAXq9nt3G398fY8aMwY8//ojo6GjIZDLIZNXT9Pbv3x+bNm3C9OnTMWTIEMycORPB\nwcG4evVqo/tbLBZj+/btOHXqFLKzs3HlyhVoNBoAQExMDObOnYu8vDwMHjwYCxYsgEQiqXd5Y3z3\n3Xe4ePEi9u7dC6D6C0zdRDz9+/dvVD2EtDa6SYy0S7XBqVevXli0aBGWLFnC3kQVFRWFM2fOWN3m\nzJkziIqKQlhYGAwGA27fvm1WRq/XY86cOSgsLKx339YCDcMwqKqZX14oFFpkWqtr3rx5GDFiBN54\n4w1kZWXho48+slruxRdfREVFBbZv385eNzaZTOjWrRsOHjyIQ4cO4dChQ9izZw+GDh3KbsflcsHh\ncMxOi9dNRFP31P295e5tt0AggMlkwr3TJdQ93rqj4qCgIBw9ehTz5s1DRUUFZs6ciaNHj6Jfv374\n66+/UFlZaVZPfn4+5s6dy94oVrssISEBCoUCAwYMwPPPP8+u69evH06cOIHHHnsMubm5mDp1Ks6d\nO1fv8sYwGo1455132P78/PPP2XsC7j0+QhyJAjRp9yZMmICoqCj2rtyxY8dCLBYjPT2dvWtbq9Vi\n9erVcHNzQ2xsLIRCIdLS0rB48WIUFxcDqA7Oa9asgUajgVQqrXd/bm5uiIiIYEeoKpUKhw4dYoOk\nrbl/hEIhAMDb2xtvvfUW3n77batfKAQCATZu3IidO3eyxxEREYG///6bLX/16lWMGzcOBQUF7HYS\niQRRUVHYv38/AOD27dv4+eef2fUNte/kyZNQqVQwGo34/PPPMWzYMISGhkIoFOL48eMAqgPokSNH\nMGTIEIvt9+zZg1dffRVDhgzBggULEBMTg2vXriEgIAATJ07E4sWL2VPParUaq1atgo+PD9snAHDx\n4kX4+PjgqaeewpAhQ3Dq1Cm23Rs3bsSWLVswevRoLFmyBN26dcPff/9d7/LGGDp0KD766CMwDAO9\nXo9//etf7HtLiDPRKW7S7li7C3np0qWYPHkyTp8+jSFDhmDnzp3YsmULkpKS2NPZo0aNwq5du9jR\n5Jw5c+Dq6orZs2eDw+FAp9MhOjoa27Zts7nfN998E6+99hr2798Pg8HAPvaVm5vbpLuko6KiMH/+\nfLzwwgtsQK0rNDQUr7zyCpYvXw6g+vr0e++9hzfffBM6nQ4Mw+DNN9+EXC432+7111/HkiVLsGfP\nHgQEBCA4OJhN09pQ+7p164a0tDSo1WpERUUhLS0NfD4fW7ZswZo1a/Duu+/CZDLh2WefRXR0NH79\n9Vez7RMSEnDmzBmMHz8erq6u6NSpE3udfeXKldiyZQtSU1PB5/PZ0+bPPvusWR0xMTHYv38/xo4d\nCz8/P4wePRp+fn7Izs7GzJkzsXDhQkycOBFCoRA9e/bEhAkTUFZWxi4XCATo1asXJkyYgMzMTJvv\nwdKlS7F27VpMmjQJBoMBQ4YMwT//+U+bfUWIvdFUn4Tch7Zv346xY8ciNDQUarUakyZNwo4dOxAW\nFubsphFCGolG0ITch7p06YLnn38eXC4XRqMRc+fOpeBMSDtDI2hCCCGkDaKbxAghhJA2yOGnuA0G\nI0pKKm0XJM3m7S2mPnYA6mf7oz62P+pjx5BK3Zu8jcNH0Hw+z3Yh0iLUx45B/Wx/1Mf2R33cdtEp\nbkIIIaQNogBNCCGEtEGNCtDnz5/H9OnTLZafPHkSU6dORUpKCjuPLSGEEEJazuZNYh988AEOHz4M\nNzc3s+UGgwGvv/46Dhw4ABcXF6SmpmL06NFmCQsIIYQQ0jw2R9AhISHYsmWLxfK//voLISEhkEgk\nEAgE6N+/v9X5hAkhhBDSdDZH0HFxcWxS+rrUajXc3e/eNu7m5gaVStW6rSPkHnXn1WHYZfWUtbLe\nWlFr683KMZbLGACuehMqqhir+2/tfde33vq+GYv1jd232c821jd03A3V31Db7j32YlThTonJYn3L\n33MrnyMrFdXXtrb3nluub+y+Pcq1KCs3NrBvy89Tc/Zd33qbn7dW/gy3Vl+ar7csUbduIY+DOfXn\n36lXs5+DlkgkZgnRKyoq4OHhYXO7Uq0JS38xokxb/UvX4C97E94wxso7YXsbyxc2tzHbTf1viq0/\nVI3dt9k2jdw3oG3ZHwsby+rW1NhtWrLvtkvr7AZ0ADpnN6AD0NsuQhrGGOGqK4abRgk3rRIGngty\n/YebFZkzuOnVNjpA3zsjaFhYGLKzs1FeXg6RSIQzZ85g9uzZNus5r6zC6Vt68DiAoO4Jdo7Zf2Y/\n15dQxtp6a0WtrTcrZ2O9tf1z6pTgWGl7ffuufcm1stBs37XLrO674bYJ+FwYDCabbatvPcfih3vX\nc+rdd92y5sduWbix74vVttWz/5Z8Jsx+bsRnQijkQ683NNi2pvR73VfNe1/Mt7Vec+P7pr5+d+T7\n4uoqhFajN1/I/shp0r7rrm/d34em7bu+/dvn98H276pE4oKKCp3Zsua2rb71zXpfWvA5q7dOK9ua\n/S23UlFDx1NRWojzx/dAfScfJuPdvwViTz88PTSO3UbYzEfNGx2ga9OuZWZmQqPRIDk5GYsWLcKs\nWbPAMAySk5Ph7+9vsx5jzdmquQ8IkNyDcnXYg1TqjsJCutxgb9TP9kd9bH9SqRiFhUZnN6PNMZlM\nuHOnGEplHsrLyzB48DCLMhofD/xckg9ZQABkssCaf3LIZIEQi1s+AYzDk2Uc+0uLF74tx7ORAiR1\npwBtD/RHzTGon+2P+tj+qI/vMhqN+OqrQ1Ao8pCfr4BeX332hsPhYPnytRAIBFa3qc0x35DmTPXp\n8AhZVTOC5tV3jokQQgixA4ZhUFpaAqVSge7de4DPNw+BPB4Ply9fQmVlBaRSf7MRMaee6wSNCc7N\n5fAAXXuKm09zmBFCCLGzixfPITv7byiVeVAqFdBqNQCAZ555AXJ5J4vyaWnPwNPTyyJ4O4Pjs1mZ\nqs+o0wiaEEJISzEMA7VaBaFQCBcXkcX6M2f+hxs3/gSHw4Gvrx+6desOmawTXF3drNQG+Pr62bvJ\njea0ETSPRtCEEEKaqLi4CLduZdeMiKv/VVRUIDn5cURERFmUj4sbB2AcAgJkEApdHN/gFnD8CJqh\nETQhhJCGmUwmcLmWI7mff/4vfvnlv+xrb28fdO4cCjc3idV6goND7NZGe3N8gK65m5/HpQhNCCEd\nnclkQnFxEZTKPCgUeez/UVEPIS7uEYvyfftGwM9Pyt7AJRJZnta+Xzj+FHfNQ118is+EENLhZWX9\nisOH95kt8/DwrPfu6C5dQtGlS6gjmuZ0TrgGXXOKm65BE0LIfclkMqGk5A6USgV7ndjDwxMTJyZZ\nlA0O7ozIyAFmjzTdmz2xo3Lac9A0giaEkPuPUpmHf/97C/R683nUO3UKtlpeJgvElCkpjmhau2Mz\nQDMMg5UrV+Lq1asQCoVIT09HcPDdjj527Bi2b98OLpeLpKQkpKamNlgf3cVNCCHtD8MwKCsrY0fE\narUa8fEJFuW8vHzg5eXFjojl8k6QyeRwd7edTImYsxmgjx8/Dr1ej4yMDJw/fx7r1q3D1q1b2fXr\n1q3D4cOHIRKJMGHCBMTHx5ulobwXPQdNCCHth8FgwMcf74BSmQeNRsMu53A4GDPmEYtHl0QiEebP\nf9nRzbwv2QzQWVlZiImJAQBERETg0qVLZusFAgHKysrYadDqmw6tFnuTGN3FTQghTldWVoY///wT\nSmUeBg4cajHfNJ/Px507xRCL3dC1a3fIZIGQy6uvFQsEQie1umOwGaDVarXZiJjP55s9nzZr1ixM\nmTIFYrEYcXFxkEisP4tWi0bQhBDiXN9/fxI3b16HQpGHigo1u7xr125WrxW/8MKrbWLqy47GZo9L\nJBJUVFSwr+sGZ4VCgd27d+PkyZMQi8V46aWXcOTIEYwdO7be+mrSFMPPRwyp1DIzCGkdzcmcQpqO\n+tn+qI+brqKiAjk5OZDL5fDwsLz2m5v7N65fvwZfX1906xaGoKAgBAUFoXv3ELqDug2xGaCjoqJw\n6tQpjBs3DufOnUN4eDi7TqfTgcfjQSgUgsPhwMfHB+Xl5Q3WV3uTWHlZJQpBd4rZA6WPcwzqZ/uj\nPm6cmzf/wvXr19iEEGVlpQCAKVNSEBk5wKL8I48kIDExFa6urmZ9XFlpQmUl9bc92CXdZFxcHE6f\nPo2UlOrb4NetW4fMzExoNBokJycjISEBKSkpEIlE6Ny5MxITExusj05xE0JI0+l0WhgMRqsj3D/+\n+B2nT/8AAHB390B4eE/IZIEICJBbrcvb28eubSWtg8MwNZNjO8jSE+U49IcWn413QaCERtD2QKMO\nx6B+tr+O2sdqtcosIYRCoUBJSTFiYkZg7Nh4i/IFBUqUl5dDLg+sd07q+nTUPnY0u4ygW1vtXdw0\ngiaEdHT1JYT444/LOHRoL/taLBaja9du8PGxngrR318Gf3+Z3dpJnMN5+aDpMStCSAfBMAxUqnI2\nGUTtFJh+flJMm/akRfnQ0DDExT1S80hTINzdPWw+wkruP07LB01TfRJCOorc3NvYvv1ds2UuLiIE\nBFgf9fr6+mH48NGOaBppw5w4gnb0ngkhpHWp1Sp2NKxQ5EGjqcSMGf+0KOfvH4DevfvVTH0ZCJks\nEF5e3jQqJg1yQoCu2TF9Lgkh7VRVVRXeemsdVCrzx0qFQiH0ej2EQuE9y13w+OMzHdlEch9w3k1i\nNIImhLRBGk1lnVGxAhMmTIaLi/l80wKBAD4+vggM7MReJ5bJAuHj42v1pi9CmsPxI2gjPQdNCGl7\nDh3aiz//vMpO8lFrwIBodO7cxaJ8WtozDmoZ6aicNoKmm7gJIY6i0+mQn189Ku7WrQd8fHwtyqhU\n5TAajejWLdxsVOznJ3VCiwlphXzQFy5cwPr16wEAAQEBWL9+vUU2lLoMpurRM90cQQixpwsXzuL3\n3y9AqVTgzp1i1M7JlJCQbDVAp6TMaPBvFyGO1uJ80MuXL8d7772H4OBg7N27Fzk5OQgNDa23PoOJ\noevPhJAWq6qqQkGBEi4uIqujXIUiD7//fhGurmJ06dKVHRF37drNan0UnElb06J80Ddv3oSXlxd2\n7dqFP//8EyNGjGgwOAPVz0HTHdyEkKYqKMjHH3/8zk72UVRUCIZhMHjwMIwfP8mi/KBBQzFw4BB4\neHjSGTvSLrUoH3RJSQnOnTuHFStWIDg4GHPnzkXfvn3x8MMP11sfjaAJIfUxGo3QaCohkVjOW5yb\nextHj34NAHBxcUFwcAjk8kCEhVkfEXt4eNq1rYTYW4vyQXt5eaFz587sqDkmJgaXLl1qOEAzgIDH\noRyvdkb96xjUz82n0+lw48YN5OTksP8UCgV69OiB5557ji1X28fR0ZGQSr0QFBQEX196nKk10ee4\nbWpRPujg4GBUVlbi9u3bCA4ORlZWFqZOndpgfUYTwAVD2VPsiLLTOAb1c+MwDGP1FHNubg62bdvE\nvhYIBJDLA+HrG8D2q3kf8xAUVD1aLi6uuLc60kz0OXYMp+SDTk9Px4svvggAiIyMxPDhwxusz2Bi\nwKPrQYTcl7RajdkkH0plHnQ6LZ5/fqFFWX//AAwbNoq9ecvX149GxYTUYTNAczgcrFq1ymxZ3RvB\nHn74Yezdu/fezeplNNEsYoTcj6qqqpCevhx1U8zzeDz4+wegqqrK4i5pgUCAMWPGO7qZhLQbDp+o\npMrEQOjwvRJCmkuv1yE/X2mWKnHGjNkQiVzNygkEAkRGDoBYLIZMVj0qlkr9wePxnNRyQto3p6Sb\n5NMImpB2Ydeu93HjxnWzUTGXy0VxcRE6dQq2KJ+U9Jgjm0fIfc0pU33SPNyEOJfBYEBBQT4Uilwo\nlQoMGPCw1dzEHh6eCAkJZa8Ty2SB8PcPoEk9CHEAp+SDppvECHGO06e/x2+/nUFhYQFMJhO7XCqV\nWg3QU6akOLJ5hJA6nJIPmk+ZMghpdUajEUVFBVAoFPDz80NQUGeLMpWVlSgpuYOgoM6QyeRsUghr\nwZkQ4lxOuQZNp7gJaR1//30TWVn/g1KpQEGBEkajEQAwcOBQqwF6xIhYjB49lh5nIqQdcHiAZkCP\nWRHSWCaTCXfuFKOqqgpyeaDF+rKyEpw9+//A5/MRECBnrxV36WJ9Tny6dkxI++GUB55oBE2IdWq1\nCpcuXWAn+8jPV6Cqqgpdu3bDrFnzLMqHh/fE/PkvwddXSo8zEXKfaXE+6FrLly+Hl5cXO6tYQ2gE\nTToyhmFQUaG2mhBCpVIhM/MggOpJPqRSf8hkgQgJsT4idnUVw9VVbNf2EkKco8X5oAEgIyMD165d\nQ3R0dON2SiNo0kGYTCbk5eVAqVSwjzQplQoIBAK8+uoKi/JSqT+mTElhJ/ng82lWH0I6qhblgwaA\ns2fP4uLFi0hJScGNGzcatVMaQZP7TX0JIYxGI/79783sI00cDgd+flLIZIEwGAwWAZjP5yMycoBD\n2kwIadtalA+6sLAQmzdvxtatW/H11183eqf0HDRpzwwGAwoL83H9egmuXbtRM/1lHp5//lWIxean\nmwUCAUaOHAN3d3fIZHL4+8sgFAqd1HJCSHvSonzQ3377LUpLS5GWlobCwkLodDp07doVCQkJDe+U\nRtCkHdu69W0UFOSbLfPx8YVKVWYRoAFg5MhYRzWNEHIfaVE+6OnTp2P69OkAgIMHD+LmzZs2gzMA\nuLkKKEG4nVH/No3RaER+fj5ycnLYfxMnTjTL3Farf/8oqNVqBAUFITg4GJ06dYJIJHJCqzsG+izb\nH/Vx29TifNDNUaWvogThdkQJ2Jvmm2/+D//732kYDAaz5d2794ZE4mdRfsiQ0QDu9rNKVQWVqsoh\nbe1o6LNsf9THjtGcL0EtzgddKzExsdE7peegiSPUTvJRmyIxJCQU3bv3sCgnFrvB3z+ATQZROwWm\ntdPVhBDiKDRRCbnvXLnyO3744STy8xXQ6/Xs8ujoQVYD9PDhozB8+ChHNpEQQmxySoCmZBmkuRiG\nQWlpCZRKBfh8vtWAazQakJt7m53ko3ZEHBjYyQktJoSQ5qERNGnz7twpxk8//cieqtZqNQCALl26\nWg3QPXr0xrJl6TTvNCGkXXNOgKbHrEgdDMNArVahrKzUagamqqoq/PLLf8HhcODr64du3cIhkwUi\nKMhyylmAEkIQQu4PzjnFTSPoDq2qqqomIUQe+6+iogJisRiLFq2ymJHLz0+KuXOfRUCADEKhi5Na\nTQghjkUjaGI3lZUVEIvdLJZzOBwcOvQFm7vY29sHnTuHQiaTw2g0Wkx/yePxEBwc4pA2E0JIW0Ej\naNIqCgsL6iSDyINCkQeVqhyvvrrCImsTn89HUtJj8PT0hkwmg0jk6qRWE0JI20UjaNIq/vOfj1BY\nWMC+9vDwRHh4T+h0OqtpFSMiohzZPEIIaXecEqC5lCyjzTOZTDWPM1WPhmv/T0p6DF27drMoP2jQ\nUFRVGSCTySGXB1o9tU0IIaTxbAZohmGwcuVKXL16FUKhEOnp6QgOvnv3bGZmJj755BPw+XyEh4dj\n5bR82rYAACAASURBVMqVtndKI+g2b//+DJw//5vZMjc3CTSaSqvlo6MHO6JZhBDSYdgM0MePH4de\nr0dGRgbOnz+PdevWYevWrQAAnU6Hd999F5mZmRAKhViwYAFOnTqFkSNHNlgnPQftHAzDoKyszOzu\n6V69+iEiItKibGhoGBiGgVx+d/pLd3cPJ7SaEEI6JpsBOisrCzExMQCAiIgIXLp0iV0nFAqRkZHB\n5rc1GAxwcbH9GAyNoB3vt9/O4JtvvoRGozFbLpG4Ww3QAwY8jAEDHnZU8wghhNzDZoBWq9Vwd797\nkw+fz2dzQnM4HPj4+AAAPv30U2g0GgwebPtUJ42gW5darWKvE7u5STB2rOW80q6urhCL3dC1azfI\nZIHsyNjT08sJLSaEkGqfffYxvvhiD/bt+z8IBAKsXbsKsbFjER09kC0zefJYHD58BADwww/fYd++\nDDAMA71ej9TUf2DEiNFN3u+XXx7El18eBJ/Px4wZszB48FCz9dev/4kNG9aBx+MhKCgYL7+8GHw+\nHxkZu3H06DdwcXFBUtKjiIsb17IOaIDNAC2RSFBRUcG+rg3OtRiGwRtvvIHs7Gxs3ry5UTv19nSF\nVEr5c1siJycH+/btQ05ODlSqu6niwsLCMHbsKIvUZlLpIAwbNsjRzbzvUR5d+6M+tj9n9vHJk0cx\nadJE/O9/3yMxMREikQCenq5mbeJyuZBK3fHbb7/h0KEvsGvXhxCJRCgtLcVjjz2GqKh+CAsLa/Q+\ni4qKcOjQXhw8eBBarRapqakYPz7WbBbCp55ai+XLlyEiIgKbNm3C0aNfYvDgwThx4ggOHNgPhmGQ\nlJSEsWNHwdfXt1X7pJbNAB0VFYVTp05h3LhxOHfuHMLDw83WL1u2DCKRiL0u3RiVai0KCyl/bkMq\nKytrZthSo1+/By3Wl5frcOXKFXh5eaNnzz7s3dNyeXVCCMrvan+UR9f+qI/t7+NrHHxzTWO7YBOM\nCOZhXoTtKXfPns2CTBaIMWMm4rXXlmHo0FhotVUoK9OYve8mE4PCQhU+/fQ/SEx8tE4Odh62b/8I\nEonErPz69WuQm5vDvvbw8MCaNW+wr//73/+hd+8HUFJSfdxyeSf88stZ9OzZiy2Tl5eHwMCuKCxU\nISysFw4fPgBXVw/06/cgSku1AIDOnbvghx9+xtChw20eq13yQcfFxeH06dNISUkBAKxbtw6ZmZnQ\naDTo06cPDhw4gP79+2P69OngcDiYMWMGYmNjG6yTnoO2pNPp8MMPJ9nHmcrLywAAIpEIfftGWEx/\n6evrhyVLVsPVlSb5IIS0T5mZhxAfn4Dg4M4QCAS4fPmS1XK1f/+KiooQGBhktk4ikViUX7hwaYP7\nraysMNvO1VWMigq1WZnAwCCcP38WERGROH36R+h0WoSFdcOnn34EjUYDvV6HixcvNCo4N5fNAM3h\ncLBq1SqzZaGhoezPly9fbvJOO+o1aK1Wi4ICJYKDQywCLp/Px+nT38NgMMDd3QPdu/eouXs6ECaT\nCTwez6w8l8ul4EwIabGXhkgwM5xx+H5VKhV+/vknlJSUYt++z1FRUYH9+7+AWOxmlscdADstsEwm\nQ0GBEmFhd+diuHjxPHx8fNGp093AvX79GuTk3GZfe3p6mo2gxWI3s0u3lZWVFhMqLVq0HO+8sxFG\n4w488MCDUKtV6Ny5C5KSkrFgQXVugD59+sHLy3738dBMYnZ09epl5Obm1IyKFSgpKQYAvPLKMnh4\neJqV5fF4mD37Kfj4+MLNzfIbISGE3E+OHPkK8fGT8fTT8wEAOp0WycmTkZo6Hd9/fxJDhw4DAJw/\nfxZdulQPCsePn4T339+MyMgBEIlEKCm5g7VrVyE9/Q2zum2NoHv37oMdO7ahqqoKOp0Ot279ja5d\nza9h//TTj1ixYg08PDywadObiI4eiNLSUlRWVmDr1g9QUaHGvHmz0KdPv9bqEguUD7qFqqqqwOFw\nLBI8AMCxY99AqVQAQM0d1N0hl8vrrYsSQhBCOoqvvvoSy5a9xr52+f/t3XlAU2fWP/Bv9gABkTVB\nWQRFUBTEZbTKFBXUqq1YpcJY9KdWHZdxrF3UqmCrltqW2gWX6lQ7tR21vkVqmY5bUd8pbgyKVepo\nX3cgAWSRJSHr/f0RuZomEgETtvP5p83Nw71PTpCT5+bec0RiREePglpdD0dHR8ya9Sc4OUnA5/Px\n5purAABhYf3wwguT8eqrC8Hj8aHRaLBgwRKL1Q0b4+bmjvj4aVi4cA4YBpg3bxEEAgFu3bqJjIxv\nsWzZcvj6+mPp0gUQCkUICemDceMmgMPh4M6d25g7dwa4XB4WLlxi06qJHIZh7HpuI2xzKT4eKUS4\nJ8/64DampqYaxcVFD4p8GJtC3LtXhpkz56Jnz2Cz8f/9bwE4HC5kMh84O7uYnda2Fbqwxj4ozrZH\nMbY9irF92OQiMVtoryvoH388iEuX8tnHYrEYfn4Bj028ISF97TU1QgghHUwrJei2k6Fra2tMWiQq\nFHIMHjwUQ4cONxvbr18EPD292CIfrq5d7bYqJoQQ0rm0Tj/oNnKR2M8/n8ChQ1km2wQCAZTKOovj\n+/QJQ58+YfaYGiGEkE6uQ57iVqmUj6yK5XB3d8ezz5qXgpPJuqF371CT0pdubu4mldIIIYSQ1tCh\nVtB37tzCvn1f4/79KpPtfn4BFhN0UFAvBAX1ss1kCCGEkBZocT/o7OxsbNmyBXw+H1OmTEF8fLzV\ngzZnBa1Wq1FSYlwVq9VqREWZt7R0cpJAr9c/KPIhYwt9eHh4Nv2AhBBCSCtqUT9onU6H9957DxkZ\nGRCJREhMTMTo0aPZDleP86SFSlQqFTIzv4VCIUdFRTka7ggTCkUYPvxZs1PRbm7uWLEi5cl2Tggh\nhLRhLeoHff36dfj7+7M1TQcOHIjc3FyMHTu20X02XMWt1WpRWqpASYkCAwYMMrsiWiQS4dq1/4LP\nFyAgIJD9nlgq9bG4X7qimhBCSEfRon7Qv3/OycnJpPWhJbKyUziUKUdFmbHIR8OqODCwJ1xdu5qM\n5XK5eP31VXB0dKLkSwghpFNpUT9oiUSC2tqHHUDq6urg4uLS6P6Oro1r0gSpF2zzUNzsg+JsexRj\n26MYt01Wvw2OjIzEyZMnAcCsH3RQUBBu376N6upqaDQa5ObmIiLCvHcxIYQQQprGai3uR6/iBoz9\noAsKCqBSqRAfH48TJ04gPT0dDMNg6tSpSExMtMvECSGEkI7M7s0yCCGEEGIdlcwihBBC2iBK0IQQ\nQkgb1CqlPglpipCQEAQHB4PL5YLD4UClUsHZ2RkpKSkICzM2L1GpVPj0009x/PhxCIVCcDgcjBw5\nEgsWLIBIJGL3deDAAezbtw9qtRparRaRkZF44403TG4XbLB161Z8++23GDZsGN59991mzX3lypXI\nycmBu7s7GIaBTqdDSEgIVqxYAQ8Pj2btMzs7G6dPn8aqVaseO2b16tWYMGEChg0b1uT9nz59Ghs3\nbgSHw0FZWRn0ej2kUikAYN68eXjuueeaNW9LGns/0tPTUVVVhdWrVz+1412+fBk7duzAJ598AoVC\ngblz54LH42HNmjX46quv8Mknnzy1YxHSYgwhbVxISAhTVVVlsu2LL75gpk2bxjAMw+h0OmbatGnM\ne++9x9TX1zMMwzD19fXM+vXrmenTpzN6vZ5hGIbZunUrM336dKa8vJz9ubfffpuZPn26xeOOHj2a\nycvLa9HcV6xYwezcudNk27Zt25hJkyYxBoOhRfu2h88++4xZt26dTfZt7f2w5bEZhmEOHDjAzJo1\ny2b7J6SlaAVN2jyGYdiCNgCg1+tRXFwMV1dXAMC//vUvMAyD5cuXs2NEIhFWrVqFuLg4HD16FH/8\n4x/x+eef4+DBg2wpWh6Ph+XLl+Po0aPQ6XTg8x/+c3j11VehUCiwatUqLFmyBJGRkUhJSUFRUREA\nIC4uDnPmzEFRURGmT5+OoKAgFBUV4euvv7a6Mp4/fz4yMjKQk5ODESNG4Pz580hLS4NKpQKXy8Xi\nxYsRHR0NAPj888+RmZkJPp+PgIAApKam4ujRozh8+DC2bduGI0eOYNu2beByueDxeHjjjTcwaNAg\nJCUlISkpCWPGjMGxY8ewefNmGAwGSCQSLF++HP3790d6ejqKiopQWlqK4uJiuLu7Y9OmTfD0bLx2\n/cqVK1FVVYXCwkJER0djyZIl+PDDD5GbmwuDwYDQ0FCsXr0aTk5OKCkpwbp16yCXy6HT6TBhwgTM\nmzcPKpWq0fdDq9WaHPP48eP4/PPPodPpUFFRgUmTJuGvf/0rlEolVq5ciTt37oDD4SAsLAzvvPPO\nY7efO3cO69atw5o1a/DJJ5+gtrYWM2fOxKJFi7Bu3Tr88MMP0Gq1j309o0aNQnh4OK5du4ZXX30V\nMTExjcaKkJagBE3ahRkzZoDD4aCiogIikQgjR45kTzvn5+dj0KBBFn9u2LBhyMvLQ/fu3eHo6GjS\n6AUwJvKJEyea/dymTZswatQopKWloU+fPkhKSkJMTAxmzpyJ2tpaTJ8+HTKZDOHh4VAoFPjoo48Q\nGRn5xK8nJCQE165dQ//+/fHWW29h586d8PHxQWlpKV566SXs3bsXBQUFyMzMxP79+yGRSLBx40Z8\n88038PLyYvfzwQcfIC0tDf3798epU6dw7tw5k1jcuHEDa9euxb59+9CtWzecOXMGCxcuxOHDhwEY\nS/lmZmbC0dERCxYswL59+7B48WKr81er1fjhhx8AAJs3bwafz0dGRgYbuw8//BApKSl48803MWvW\nLERHR0Oj0WDu3Lnw8/ODr69vk96PL7/8Eu+//z78/PxQWlqKkSNHYubMmTh58iSUSiUOHDgAg8GA\ntWvX4u7duzh//rzF7Q2GDBmCJUuWsB90zp07xz63fft2s9eTlpaG5ORkAEBwcDA2bdpkNUaEtBQl\naNIu7N69G126dMGVK1cwd+5cDBgwwKQpi06ns/hzGo0GfD4fXC4XBoOhycdlGAYqlQrnz5/Hzp07\nARgr6E2ePBn//ve/ER4eDj6f3+QCPRwOB2KxGBcuXEBZWRkWLVrEniXgcrm4evUqTp8+jXHjxrG1\n7hvOEBw4cIDdz4QJE7Bw4UJER0fjmWeewSuvvGJynDNnzmDYsGHo1q0bAGDo0KHw8PBAQUEBAGOi\ncnR0BAD06dMHVVWmrVof59EPIydOnEBNTQ1ycnIAGN8Ld3d3qFQq5Obmorq6Gh9//DEA47UCV65c\ngb+/f5Pej61bt+LEiRM4ePAgbty4we5r4MCB+Pjjj5GUlIThw4dj5syZ8PX1BYfDsbhdLpdbPdbj\nXk+Dx30YJORpowRN2oWG5BUaGoqVK1di1apViIiIgI+PDyIjI/G3v/3N4s/k5uZi0aJFCAoKgk6n\nw927d01WbRqNBosXL8aGDRsee2rXUiJhGIY9DSsUCs06q1lTUFCAl19+GTU1NejZsyf27dvHPldS\nUgJ3d3ecPn3apAZ9bW0tqqurTfazdOlSTJ06FTk5OThw4AB27NjBrvwa5sn8rtSBXq9nP9CIxWJ2\ne1Pq3Ts5OZnsb9WqVWxTHaVSCbVaDb1eDwDYt28fhEIhAKCiogIODg7g8XhW348GKpUKcXFxGDNm\nDAYNGoSpU6fi2LFjYBgG3bt3x5EjR3Du3DmcOXMGM2fORHJyMsaMGYPDhw8jNzfXZHvD1yKNedzr\nadDwgYYQW6PbrEi7M2HCBERGRrJ/xMeOHQtHR0ds2LCB/UNaX1+PdevWwcnJCTExMRAKhZg7dy7e\neustlJeXAzAmg/Xr10OlUjX6vauTkxPCw8PxzTffAABqamqQmZmJESNGAIBZAmyMwWBAeno63Nzc\nMGjQIISHh+PWrVvIzc0FAFy9ehXjxo1DWVkZhg0bhqNHj7K18D/99FPs2rWL3Zder8eoUaOgVCox\nbdo0pKSk4MaNGyZnE4YOHYpTp06hsLAQgPEK7ZKSEvTv3/+J52xNVFQUvvnmG2g0GhgMBiQnJ2PT\npk2QSCQIDw/HF198AcAYt5dffhk//fQThEIhXnnlFYvvR319vcn7cfv2bSiVSixduhTR0dE4e/Ys\ntFot9Ho99uzZgxUrVmD48OF47bXXEBUVhWvXrmHPnj1YuXKl2faWvB5C7I1W0KTNs7SyW716NSZN\nmoScnBwMHz4cO3fuxObNm/Hiiy+yp7NHjRqFXbt2gcfjATDeIuTg4IA5c+aAw+FArVZjyJAh2Lp1\nq9XjfvDBB3jnnXfw3XffQafT4YUXXkBcXByKioqsrjy//PJLHDx4EIAxQffr1w/bt28HALi5ueGz\nzz7DBx98ALVaDYZh8MEHH0Amk0Emk+HGjRtISEgAh8NBr169sG7dOvb7Yx6Ph1WrVuG1116DQCAA\nl8tFamoqBAIBO6egoCCkpKRg8eLF0Ov1cHBwwLZt29jT5k/DwoUL8f7772Py5MlgGAahoaHs6fi0\ntDS88847eP7556HT6TBx4kT2O+b58+fD0dHR7P1o6DffICQkBM8++yyee+45eHl5ITIyEn379sWd\nO3cwefJknDt3DuPHj4eDgwO6deuGmTNngs/nW9x+5cqVFr0e6qpH7IlKfRJCCCFtEJ3iJoQQQtog\nStCEEEJIG0QJmhBCCGmDKEETQgghbZDdr+LW6fSorFTa+7CdSteujhRjO6A42x7F2PYoxvbh6Wne\nkMcau6+g+XyevQ/Z6VCM7YPibHsUY9ujGLdddIqbEEIIaYMoQRNCCCFt0BMl6IsXLyIpKclse3Z2\nNqZOnYqEhATs37//qU+OEEII6aysXiT2t7/9Dd9//71JcXzA2OHlvffeQ0ZGBkQiERITEzF69GiT\nDkOEEEIIaR6rK2h/f39s3rzZbPv169fh7+8PiUQCgUCAgQMHsgX/CSGEENIyVlfQsbGxKCoqMtte\nW1sLZ+eHl407OTmhpqbG6gHHfHUPej2V/7YlHk9NMbYDirPtUYxtj2L89HD0GgiVJeAwOtS79DB5\n7qdZTb/Nqtn3QUskEtTW1rKP6+rq4OLi8kQ/y+NRRxhboxjbB8XZ9ijGtkcxbh6uphbOJf+BsE4O\nYZ0CgvpycMBALemG4vCFLd7/Eyfo3ze9CgoKwu3bt1FdXQ2xWIzc3FzMmTPH6n6OzPBAWZn1lTZp\nPk9PZ4qxHVCcbY9ibHsU48bpdDpUVlbA09PL7LnKyjqkpR0FAIjFDpAG9IBU6oPu3X0RESFq8bGf\nOEE39EHNysqCSqVCfHw8Vq5cidmzZ4NhGMTHx8PLy/wFEEIIIe2BwWDA9eu/QaEohkIhh0JRjLKy\nUvD5fKxevR5crullW66uXfHyy7MglfqgSxfXp94vvFX6QdOnNduiT8T2QXG2PYqx7XXGGOv1enC5\nXLOEajAYsH79amg0GgCAUCiEt7cMMpkPxo6dAJFI3OxjNqfUp91rcRNCCCH2olQqH6yIiyGXG1fG\nZWUlWLLkDbi5uZuM5XK5GDfueTg5OUEq9UHXrm5mq2Z7ogRNCCGkw/r737ejqKiQfczn8+HtLUV9\nfb3F8UOGDLPX1KyiBE0IIaRdUavr2e+I5XLjf2NixqFnz2CzsRERAxEY2AtSqQ9kMhnc3T3B47WP\nBiGUoAkhhLQb//rXD8jJOWmyjcvloqKi3OL4YcOi7DEtm6AETQghpNVptVqUlMjZlbGfXwD69x9g\nNs7DwxOBgT0hlfpAKjVewOXp6Q0+v+Ols473igghhLQbV64U4MiRf+LevTKTehtKpdJigh48eCgG\nDx5qzym2GkrQhBBCbEKn06GsrBRyeRF4PD7Cw80TLp/PR3V1Nfz8AiCT+bArYy8vaSvMuG2hBE0I\nIeSpqagoR3b2EbbIh16vBwDIZD4WE3RQUC+sXr3uqRf56AisJmiGYbB27VpcvXoVQqEQGzZsgK+v\nL/v80aNHsW3bNnC5XLz44otITEy06YQJIYS0Hr1ej/LyMlRVVSE4OMTseR6Ph/z8PAgEAshk3dir\np318ulvcX2veZ9zWWU3Qx44dg0ajwd69e3Hx4kWkpqZiy5Yt7POpqan4/vvvIRaLMWHCBEycONGk\nyxUhhJD2S6fTITf3NHs7U2mpAjqdDnw+H2vWbDC7ZcnFpQuWLl0ONzd3Sr4tZDVB5+XlISrKeJl6\neHg4Ll++bPK8QCDA/fv32dMTdJqCEELaF4PBgMrKCouVs3g8Ho4dOwy1uh48Hg9eXlL26mmDwWCW\noDkcDjw8PO05/Q7LaoL+fd9nPp8Pg8HAvomzZ8/GlClT4OjoiNjYWEgkEtvNlhBCSIsVFt5BcXEh\nFAo57t0rQWFhETQaNZYuXW6WXDkcDhISkuDi4gIPD692U+SjI7CaoCUSCerq6tjHjyZnuVyOr7/+\nGtnZ2XB0dMTrr7+Ow4cPY+zYsY3uszlFw0nTUIztg+JsexTj5mEYBgzDWDzNvHNnFm7cuAHA+B2w\nVCpF9+7d4ebmZDHenp6DbD5fYs5qgo6MjMTx48cxbtw45OfnIzj4YSk1tVoNHo8HoVAIDocDNzc3\nVFdXWz1oZ+ucYm+dsTtNa6A42x7F+MlotVqUlpaYNYWYMmUaQkPDzMYPGTIcERGDIZX6oE+fIFRV\nPaxLTfG2DZt0s4qNjUVOTg4SEhIAGC8Ke7QndFxcHBISEiAWi+Hn54fJkyc3feaEEEKaLSsrE3l5\nZ9nHHA4H7u4e7C1OvxcWFs7+v0AgAGC5cQRpXdQPugOiVYd9UJxtrzPHWK/Xo6yshL16WqGQIzS0\nD4YOHWE2tqDgEq5fv/bgliYfeHtLIRSKnug4nTnG9kT9oAkhpAO4cOE/yMzcb7YC9vDwsDi+b99+\n6Nu3nz2mRuyIEjQhhNiJwWBAefk99ntiR0dHjBgRbTaua1e3ByUvjUU+GspfisUO9p80aTWUoAkh\nxMZKS0uQkbEXJSUKaLVadruXl7fFBB0QEIgFC/5qxxmStogSNCGEtEBDkQ+FQo6ammoMHTrcbIyj\noyPk8mJ4eXmzq+GGFTIhj0MJmhBCmkir1eLQoR8glxejpEQOtVoNwFh1a9CgP5j1JpZInJGc/C4V\n+SBNQgmaEEJ+h2EY3L9/HyUlxejZs7dZYuXz+fjll3yo1fXw8PA0WRU/rtwxJWfSVJSgCSEEwKVL\n+bhz5zZ7S5NKpQQALFnyullvYg6Hg/nz/4IuXVwf3EdMyNNHCZoQ0mnU1tZAKBRavEf4zJkc3L59\n80FVRHcEBgZBKu322CunqSEEsbUW94P+5ZdfsHHjRgCAt7c3Nm7cSJ8oCSGtrrz8Hu7evcOWv1Qo\n5KitrUFCQpJJJa0GsbHjweVy4O0tg0j0ZEU+CLGlFveDTk5OxmeffQZfX1/s378fhYWF6NGjh00n\nTQghDR5t4POonJyTOHfuNPu4SxdXhIT0gYODo8X9BATQ3y3StrSoH/TNmzfh6uqKXbt24bfffkN0\ndDQlZ0KITRgMBlRUlD/SDMK4Kh44cAhGjRpjNj4sLByenl7sBVyPS8yEtFUt6gddWVmJ/Px8pKSk\nwNfXF/Pnz0dYWBj+8Ic/2HTShJDO59y5U8jKyjTZ5uzsAg7HfPUMAIGBPREY2NMeUyPEJlrUD9rV\n1RV+fn7sqjkqKgqXL1+mBE0IeSIMw6CqqtKkRaKrqyvGj59kNtbXNwAREQPZ25lkMh84OUlaYdaE\n2EeL+kH7+vpCqVTi7t278PX1RV5eHqZOnWr1oNSA3fYoxvZBcW6+27dvY9OmTVCpVCbbe/ToYRLX\nhv/39AxFRESoXefYWdDvcdtktd3ko1dxA8Z+0AUFBWw/6LNnz+LDDz8EAAwYMABvvfWW1YNSazPb\novZx9kFxtoxhGNTUVLPfEyuVSjz33PNm45TKOuzYsdmkyIdM5vPgtLWx2AfF2PYoxvbRnA9B1A+6\nA6J/cPZBcTal1Wqxe/cXbFJuwOVysWbNhmbdfkkxtj2KsX1QP2hCiM3U1dWyq+KhQ0eY1ZsWCAQo\nLS2BSOSAgIDAR9ol+piNJYRYR/9qCCGPdfLkT7h16wbbqalBz57BFjsxvfbaW1SoiJCnhBI0IZ2Y\nSqWCQlEMLy8pnJyczJ7/v/+7hps3r6NLF1f07h3Kroi7dOlqcX+UnAl5eihBE9KJ3Lx5Hdev/waF\nQg6FohhVVZUAgPj46QgPH2A2Pi4uHg4OjnB0pCIfhNgbJWhCOhi1Wg2DQW+xclZBwS84cyYHgLFH\nccOpak9PL4v7cnf3sOlcCSGPRwmakHastrbGrCFERUU5nn12NGJixpmNHzToD+jduw+kUhmcnV1a\nYcaEkCdFCZqQdoBhGPbe4EcVFFzCDz9ksI8dHBwREBAIV1fL3xEbr6y22TQJIU8RJWhC2hCGYVBb\nW2PSDMJ4EZc3EhJmmI0PDOyJmJhx7MVbLi5dLCZyQkj7QwmakDbkzp1b2LFjs8k2oVAEb2+ZxfGe\nnl6Ijo6xx9QIIXZmNUE/WupTKBRiw4YN8PX1NRuXnJwMV1dXLFu2zCYTJaQ9Uyrr2FWxXF4MtVqN\n6dP/n9k4Ly8pQkPDIJXKIJMZC324una12O+YENKxWU3Qx44dg0ajwd69e3Hx4kWkpqZiy5YtJmP2\n7t2La9euYciQITabKCHtkUajxscfv4/q6vsm24VCEXQ6nVmFLQcHB4uJmxDS+VhN0Hl5eYiKigIA\nhIeH4/LlyybPX7hwAZcuXUJCQgJu3Lhhm1kS0gapVKoHVbaM3xVPmBBnVqhDKBTB2dnFpBmEVOoD\nd3cPWhUTQhplNUHX1tbC2flhkW8+n8/2hC4rK0N6ejq2bNmCH3/80aYTJaStOHDgW1y//htb5KPB\nkCHD4OPT3Wz8ggV/tdfUCCEdiNUELZFIUFdXxz5uSM4AcOjQIVRVVWHu3LkoKyuDWq1GYGAg4uLi\nGt0n9R61PYpx86jVahQXF+Pu3bsICwuDm5ub2RilsgZ6vQ6hoaHo3r07unfvDl9fX0ilUvB4Kak+\nUQAADsdJREFUvFaYdcdGv8u2RzFum6y2mzxy5AiOHz+O1NRU5OfnY8uWLdi+fbvZuAMHDuDmzZtP\ndJEYtTazLWof1zQXL57HlSuXoVDIUV5+Dw3/JKZMScSAAQPNxms0GggEAnh5uVCcbYx+l22PYmwf\nNmk3GRsbi5ycHCQkJAAAUlNTkZWVBZVKhfj4+KbPkhA70+l0KC0tgVgshpubu9nzRUV3cfnyLxCL\nxfD37/GgmIcMAQE9LO5PKBTaesqEEGJ9BW0L9GnNtjr7J+LS0hJcvforW+SjrKwUBoMBf/zjKIwZ\nM95sfFVVFRjGAFfXrk0q8tHZ42wPFGPboxjbh01W0IS0RXq9HvX19RZbJN65cwuHD/8TgHG1262b\nL2Qyn8euiF1dXW06V0IIaQ5K0KTN02jUKCy8y66I5fJilJYq0KtXb7z88myz8T179kZi4gxIpT7o\n2tWNbmcihLRLlKBJm/G4hhAKhRw7d25jH/P5fHh7y+DlZbnrg6urK62KCSHtHiVo0irU6vpHVsTG\n/+p0Wixe/JrZWG9vGaKiRj4o9CGDu7sn3c5ECOnwKEETu1Or67F+/Ro8en0il8uFl5fUYvlLkUiE\nsWMn2HuahBDSqihBk6dGq9WipETOrowVCjlmzHjF7LYkkUiMfv0iIJE4s00hPD29zRIzIYR0ZvQX\nkTwVX3yxFbdu3TBZFXM4HJSX34NM5mM2/qWXpttzeoQQ0u5QgiaN0ul0KCsrhVxeBIVCjsGDh8LT\n08tsnETiDD+/ALYZhFRqvIiLinoQQkjztLgfdFZWFr766ivw+XwEBwdj7dq1tpwvsZOffz6B/Pw8\nlJWVQq/Xs9u9vaUWE/S0aS/bc3qEENLhtagftFqtxqeffoqsrCwIhUK89tprOH78OEaOHGnziZPm\n0+v1KC8vg1wuh6enF3x8upmNqa2teXB6uht79bRxZWx+upoQQsjT16J+0EKhEHv37mVPY+p0OohE\nIhtNlbTErVs3cP78f6BQGIt86HQ6AMCIEc9aTNAjR47BmDETqMgHIYS0Eqt/fR/XDxowXgTU0I5v\n9+7dUKlUeOaZZ2w0VdIYg8GA8vJ7KClRWHy+oqIc58+fQ0mJHJ6e3oiMHIzx4ychPDzS4niRSETJ\nmRBic99883dMmjQOWq0WAPDuu2/j3LkzJmMmTRrL/v///u8JLFnyZ/zlL/Mxf/4snDjxU7OPXVlZ\nicTEF9ljP6qoqBALF76CxYvnIS1tI7v94MEDeOWVGfjzn2fj1Kmfm33sJ9GiftCA8Tvq999/H7dv\n30Z6evoTHZR6j7bc/fv3kZ+fj8LCQhQWFqKoqAhqtRphYWEIC+tlFuPhw4cgLKw39Sx+yuh32fYo\nxrbXmjHOzj6CF154HmfPnsTkyZMhFgvQpYuDyZy4XC48PZ1x/vx5ZGZ+i127voBYLEZVVRWmTZuG\nyMh+CAoKatJxf/75Z6SlpaGqqhIeHhKzC1qTkz/F8uVvYNCgQUhJScHFi2cRERGBzMz9OHDgAOrr\n65GYmIjx42MgEAieSix+z2qCjoyMxPHjxzFu3Djk5+cjODjY5Pk1a9ZALBaz30s/Ceqc8mQYhoFS\nqbTYEKKwsBD/+Mc/ADT88npBKvVBQEAgAMsxFgpdUFGhtO2kOxHqAmR7FGPb+/s1Dv51TfVU9xnt\ny8Ofw60nrQsX8iCV+mDMmOfxzjtrMGJEDOrrtbh/X2XyvhsMDMrKarB79z8wefJLqKnRoqZGC4CH\nbdu+hEQiMRm/ceN6FBUVso9dXFywfv37JseuqVHjww/TMWdOEu7dqzVLspcuXYa/f2+UldUgImIw\nfvrpBGpq1OjTpz8qK43xksm64cyZCwgJCbX6Wu3eD7pv377IyMjAwIEDkZSUBA6HgxkzZiAmJqbJ\nE+ns9Hq9STOIhkIfIpEIb7yx2my8t7cML744DVKpD7y8qMgHIaT9ycrKxMSJcfD19YNAIMCvv162\nOK6hRv+9e/fg49Pd5DmJRGI2fvly87+Zvzdo0JAH/2e947KjoxPq6uqgVCpNjufg4Ii6ulqrP99c\nVv+qczgcvP322ybbevR42Lbv119/ffqz6oR0Oi22bv2YfczhcODu7gGp1Ad6vd7stLRAIEBk5GB7\nT5MQ0sG8PlyCmcHWk9TTVlNTg9OnT6Gysgr/8z/7UFdXh++++xaOjk7QaDQmYxtu9ZRKpSgtVSAo\nqCf73KVLF+Hm5o5u3R4m7o0b16Ow8C77uEuXLmYr6Ics94B/tHGPUlkHZ2dnODk5mXzla0zYtvt6\ngJZdNqTT6XDvXinbDEKhKEZJiQKvvrocIpHYZKxIJEZ0dAy6dHGFVCqDt7cUQiFdEU8I6ZgOH/4n\nJk6chIULlwAw1uiPj5+ExMQknDyZjREj/ggAuHjxAtvLffz4F/D55+kYMGAQxGIxKisr8O67b2PD\nBtPk+yQr6IcsfzgJDu6N/PzziIiIxJkzpxAZORihoX2wffsWaLVaqNVq3LlzC4GBTfvuuykoQdvQ\nZ599iPLyeybbunZ1Q3V1NTw9xWbjY2LG2WtqhBDSqv75z4NYs+Yd9rFxkTIKanU9HB0dMWvWn+Dk\nJAGfz8ebb64CAISF9cMLL0zGq68uBI/Hh0ajwYIFSxAY2PNxh3kCD1fKt27dREbGt1i2bDkWLVqK\njRvXQ6/Xwd+/B0aOHA0Oh4P4+GlYuHAOGAaYN2+RzS4QAwAO82jxZDtpzxd9NNzO9Oh3xbGxz0Em\nM7+X+NAh43f1xiIf3SCVSiEWO9h8jnRhjX1QnG2PYmx7FGP7sMlFYuShH3/8Hrm5Z8zumevXL8Ji\ngh43bqK9pkYIIaSDoQQN46q4qqqSXRUHBAQiKKiX2TiRSAx3d0+2RWJD6UtLt0ERQgghLdGpE/Sv\nv17Czz+fREmJHGq1mt0+bFiUxQQ9evRYjB491mw7IYQQ8rR12ATNMAzu378PhaIYAoHAYsLVaDQo\nLLwDDw9PtkWiVOpjsTY1IYQQYk8dKkGXl9/D2bM5Dy7ekkOlMlbNCgoKtpig+/btj759+9v0KjxC\nCCGkOdpdgq6trUF19X2zajKA8T66U6f+/aCJhzsCA4MglfrA19ff4r4oMRNCCGmrrCZohmGwdu1a\nXL16FUKhEBs2bICvry/7fHZ2NrZs2QI+n48pU6YgPj7+qU1Oq9WioOASW+RDoZCjtrYGzs4uWL48\n2Wy8l5cU8+Ythre3jNpeEkIIadesJuhjx45Bo9Fg7969uHjxIlJTU9nGGDqdDu+99x4yMjIgEomQ\nmJiI0aNHsy0on5RKpYKDg/n9wQzD4Lvv9qDhVu0uXVwREtIXUqnMrKsWYGyF6ecX0KRjE0IIIW2R\n1QSdl5eHqKgoAEB4eDguX35YzPz69evw9/dni4cPHDgQubm5GDv28Vc6l5SU4NdffzNpCHH/fhVW\nrVpnlqSFQiEmT34JXbu6QSqVwcHBsVkvkhBCCGlvrCbo2tpaODs/rIDC5/PZ1evvn3NyckJNTeMV\nadLT01FaWso+dnZ2Qa9evVFfb3kVTQ0hCCGEdEZWE7REIjHp3vHoqWWJRILa2oetturq6uDi4tLo\n/tatW9fcuZImoCb39kFxtj2Kse1RjNsmrrUBkZGROHnyJAAgPz8fwcHB7HNBQUG4ffs2qqurodFo\nkJubi4iICNvNlhBCCOkkrDbLePQqbgBITU1FQUEBVCoV4uPjceLECaSnp4NhGEydOhWJiYl2mTgh\nhBDSkbVKNytCCCGENM7qKW5CCCGE2B8laEIIIaQNogRNCCGEtEE2S9AMwyAlJQUJCQmYMWMG7t69\na/J8dnY2pk6dioSEBOzfv99W0+jQrMU4KysLL730Ev70pz9h7dq1rTPJds5ajBskJyfjo48+svPs\nOgZrMf7ll18wffp0TJ8+HcuWLYNWq22lmbZv1uJ89OhRtlzznj17WmmWHcPFixeRlJRktr3JeY+x\nkSNHjjArVqxgGIZh8vPzmQULFrDPabVaJjY2lqmpqWE0Gg0zZcoUpry83FZT6bAai3F9fT0TGxvL\nqNVqhmEYZtmyZUx2dnarzLM9ayzGDfbs2cNMmzaNSUtLs/f0OgRrMZ40aRJz584dhmEY5ttvv2Vu\n3Lhh9zl2BNbiPHLkSKa6uprRaDRMbGwsU11d3RrTbPd27NjBTJw4kZk2bZrJ9ubkPZutoJ+0RKhA\nIGBLhJKmaSzGQqEQe/fuhVAoBGCsm04NRJqusRgDwIULF3Dp0iUkJCS0xvQ6hMZifPPmTbi6umLX\nrl1ISkpCdXU1evTo0VpTbdes/S4LBALcv38farUaAMDhcOw+x47A398fmzdvNtvenLxnswT9uBKh\nlp57khKhxFxjMTa23DQ2Ldm9ezdUKhWeeeaZVplne9ZYjMvKypCeno7k5GS2oQtpusZiXFlZifz8\nfCQlJWHXrl04deoUzp4921pTbdcaizMAzJ49G1OmTMHzzz+P6OhotscCaZrY2FjweDyz7c3JezZL\n0E+7RCgx11iMAeN3Ths3bsTp06eRnp7eGlNs9xqL8aFDh1BVVYW5c+di+/btyMrKQmZmZmtNtd1q\nLMaurq7w8/NDjx49wOfzERUVZbbyI0+msTjL5XJ8/fXXyM7ORnZ2NsrLy3H48OHWmmqH1Jy8Z7ME\nTSVCba+xGAPAmjVroNVqsWXLFvZUN2maxmKclJSE7777Dl999RXmzZuHiRMnIi4urrWm2m41FmNf\nX18olUr2gqa8vDz07NmzVebZ3jUWZ7VaDR6PB6FQyJ59q66ubq2pdgi/P6vWnLxntVlGc8XGxiIn\nJ4f9bi41NRVZWVlsidCVK1di9uzZYBgG8fHx8PLystVUOqzGYty3b19kZGRg4MCBSEpKAofDwYwZ\nMxATE9PKs25frP0ek5azFuMNGzZg2bJlAIABAwbg2Wefbc3ptlvW4hwXF4eEhASIxWL4+flh8uTJ\nrTzj9q3hO/yW5D0q9UkIIYS0QVSohBBCCGmDKEETQgghbRAlaEIIIaQNogRNCCGEtEGUoAkhhJA2\niBI0IYQQ0gZRgiaEEELaIErQhBBCSBv0/wHdNdlHbeaMIQAAAABJRU5ErkJggg==\n", "text/plain": [ - "" + "" ] }, "metadata": {}, @@ -241,15 +249,20 @@ } ], "source": [ - "yb.rocplot_compare(\n", - " [oclm, ocknn], [oclm_data.y_test, ocknn_data.y_test], \n", - " [oclm.predict(oclm_data.X_test), ocknn.predict(ocknn_data.X_test)]\n", - ")" + "feats = ['temp', 'humid', 'light', 'co2', 'hratio']\n", + "\n", + "models = [\n", + " build_model(occupancy, feats, 'occupied', LinearSVC()),\n", + " build_model(occupancy, feats, 'occupied', KNeighborsClassifier()),\n", + " build_model(occupancy, feats, 'occupied', DecisionTreeClassifier())\n", + "]\n", + "\n", + "yb.rocplot([model[0] for model in models], occupancy[feats], occupancy['occupied'])\n" ] }, { "cell_type": "code", - "execution_count": 9, + "execution_count": null, "metadata": { "collapsed": false }, @@ -265,7 +278,45 @@ } ], "source": [ - "print ocknn.steps[-1][1]" + "print(ocknn.steps[-1][1])" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "from sklearn.linear_model import Ridge, RANSACRegressor \n", + "from sklearn.svm import SVR\n", + "\n", + "feats = [\n", + " 'cement', 'slag', 'ash', 'water', 'splast', 'coarse', 'fine', 'age',\n", + "]\n", + "\n", + "models = [\n", + " build_model(concrete, feats, 'strength', Ridge()),\n", + " build_model(concrete, feats, 'strength', SVR()),\n", + " build_model(concrete, feats, 'strength', RANSACRegressor())\n", + "]\n", + "\n", + "yb.peplot([model[0] for model in models], concrete[feats], concrete['strength'])\n" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "from sklearn.linear_model import LinearRegression \n", + "\n", + "models = [Ridge(), LinearRegression(), SVR()]\n", + "yb.residuals_plot(models, concrete[feats], concrete['strength'])" ] } ], diff --git a/examples/figures/pipeline_prototype.png b/examples/figures/pipeline_prototype.png new file mode 100644 index 000000000..bffa05fc3 Binary files /dev/null and b/examples/figures/pipeline_prototype.png differ diff --git a/examples/pipeline.ipynb b/examples/pipeline.ipynb new file mode 100644 index 000000000..a2c8e5b35 --- /dev/null +++ b/examples/pipeline.ipynb @@ -0,0 +1,179 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Visual Pipelines \n", + "\n", + "This notebook demonstrates a proof of concept for a visual pipeline for analytics. \n", + "\n", + "![Yellowbrick Prototype Pipeline Objects](figures/pipeline_prototype.png)" + ] + }, + { + "cell_type": "code", + "execution_count": 14, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "%matplotlib inline\n", + "\n", + "import os\n", + "import sys \n", + "\n", + "# Modify the path \n", + "sys.path.append(\"..\")\n", + "\n", + "import pandas as pd\n", + "import yellowbrick as yb \n", + "import matplotlib as mpl \n", + "import matplotlib.pyplot as plt " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Load Datasets \n", + "\n", + "Note that if datasets do not exist, please see the `download.py` located in this directory. " + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "FIXTURES = os.path.join(os.getcwd(), \"data\")\n", + "credit = pd.read_excel(os.path.join(FIXTURES, \"credit.xls\"), header=1)\n", + "concrete = pd.read_excel(os.path.join(FIXTURES, \"concrete.xls\"))\n", + "occupancy = pd.read_csv(os.path.join('data','occupancy','datatraining.txt'))" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "# Rename the columns of the datasets for ease of use. \n", + "credit.columns = [\n", + " 'id', 'limit', 'sex', 'edu', 'married', 'age', 'apr_delay', 'may_delay',\n", + " 'jun_delay', 'jul_delay', 'aug_delay', 'sep_delay', 'apr_bill', 'may_bill',\n", + " 'jun_bill', 'jul_bill', 'aug_bill', 'sep_bill', 'apr_pay', 'may_pay', 'jun_pay',\n", + " 'jul_pay', 'aug_pay', 'sep_pay', 'default'\n", + "]\n", + "\n", + "concrete.columns = [\n", + " 'cement', 'slag', 'ash', 'water', 'splast',\n", + " 'coarse', 'fine', 'age', 'strength'\n", + "]\n", + "\n", + "occupancy.columns = [\n", + " 'date', 'temp', 'humid', 'light', 'co2', 'hratio', 'occupied'\n", + "]\n" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/plain": [ + "[('scale', StandardScaler(copy=True, with_mean=True, with_std=True)),\n", + " ('model', LinearSVC(C=1.0, class_weight=None, dual=True, fit_intercept=True,\n", + " intercept_scaling=1, loss='squared_hinge', max_iter=1000,\n", + " multi_class='ovr', penalty='l2', random_state=None, tol=0.0001,\n", + " verbose=0))]" + ] + }, + "execution_count": 12, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "from sklearn.svm import LinearSVC\n", + "from sklearn.preprocessing import StandardScaler \n", + "\n", + "model = Pipeline([\n", + " ('scale', StandardScaler()), \n", + " ('model', LinearSVC())\n", + "])\n", + "\n", + "model.steps" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Evaluation Visualization Prototype" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "from sklearn.pipeline import Pipeline\n", + "from sklearn.base import BaseEstimator, TransformerMixin\n", + "\n", + "\n", + "class VisualPipeline(Pipeline):\n", + " \n", + " def draw(self):\n", + " \"\"\"\n", + " Calls the draw method on every visual transformer/estimator \n", + " \"\"\"\n", + " for name, estimator in self.steps:\n", + " try:\n", + " estimator.draw()\n", + " except AttributeError:\n", + " continue \n", + "\n", + " \n", + "class ClassifierEvaluation(object):\n", + " \n", + " def draw(self):\n", + " yb.crplot()" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 2", + "language": "python", + "name": "python2" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 2 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython2", + "version": "2.7.11" + } + }, + "nbformat": 4, + "nbformat_minor": 0 +} diff --git a/examples/rank2d.ipynb b/examples/rank2d.ipynb new file mode 100644 index 000000000..3d12f322c --- /dev/null +++ b/examples/rank2d.ipynb @@ -0,0 +1,271 @@ +{ + "cells": [ + { + "cell_type": "code", + "execution_count": 1, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "from __future__ import print_function\n", + "%matplotlib inline " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Sliders Example \n", + "\n", + "This is an example of interactive iPython workbook that uses widgets to meaningfully interact with visualization. " + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "# Imports \n", + "import numpy as np\n", + "import pandas as pd\n", + "import seaborn as sns\n", + "import matplotlib.pyplot as plt\n", + "\n", + "from collections import OrderedDict\n", + "from sklearn.pipeline import Pipeline\n", + "from sklearn.preprocessing import Imputer \n", + "from sklearn.linear_model import LinearRegression\n", + "from sklearn.metrics import mean_squared_error as mse\n", + "from ipywidgets import interact, interactive, fixed\n", + "\n", + "import ipywidgets as widgets" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "# Data Loading \n", + "columns = OrderedDict([\n", + " (\"DAY\", \"the day of data collection\"),\n", + " (\"Q-E\", \"input flow to plant\"),\n", + " (\"ZN-E\", \"input Zinc to plant\"),\n", + " (\"PH-E\", \"input pH to plant\"),\n", + " (\"DBO-E\", \"input Biological demand of oxygen to plant\"),\n", + " (\"DQO-E\", \"input chemical demand of oxygen to plant\"),\n", + " (\"SS-E\", \"input suspended solids to plant\"),\n", + " (\"SSV-E\", \"input volatile supended solids to plant\"),\n", + " (\"SED-E\", \"input sediments to plant\"),\n", + " (\"COND-E\", \"input conductivity to plant\"),\n", + " (\"PH-P\", \"input pH to primary settler\"),\n", + " (\"DBO-P\", \"input Biological demand of oxygen to primary settler\"),\n", + " (\"SS-P\", \"input suspended solids to primary settler\"),\n", + " (\"SSV-P\", \"input volatile supended solids to primary settler\"),\n", + " (\"SED-P\", \"input sediments to primary settler\"),\n", + " (\"COND-P\", \"input conductivity to primary settler\"),\n", + " (\"PH-D\", \"input pH to secondary settler\"),\n", + " (\"DBO-D\", \"input Biological demand of oxygen to secondary settler\"),\n", + " (\"DQO-D\", \"input chemical demand of oxygen to secondary settler\"),\n", + " (\"SS-D\", \"input suspended solids to secondary settler\"),\n", + " (\"SSV-D\", \"input volatile supended solids to secondary settler\"),\n", + " (\"SED-D\", \"input sediments to secondary settler\"),\n", + " (\"COND-S\", \"input conductivity to secondary settler\"),\n", + " (\"PH-S\", \"output pH\"),\n", + " (\"DBO-S\", \"output Biological demand of oxygen\"),\n", + " (\"DQO-S\", \"output chemical demand of oxygen\"),\n", + " (\"SS-S\", \"output suspended solids\"),\n", + " (\"SSV-S\", \"output volatile supended solids\"),\n", + " (\"SED-S\", \"output sediments\"),\n", + " (\"COND-\", \"output conductivity\"),\n", + " (\"RD-DB-P\", \"performance input Biological demand of oxygen in primary settler\"),\n", + " (\"RD-SSP\", \"performance input suspended solids to primary settler\"),\n", + " (\"RD-SE-P\", \"performance input sediments to primary settler\"),\n", + " (\"RD-DB-S\", \"performance input Biological demand of oxygen to secondary settler\"),\n", + " (\"RD-DQ-S\", \"performance input chemical demand of oxygen to secondary settler\"),\n", + " (\"RD-DB-G\", \"global performance input Biological demand of oxygen\"),\n", + " (\"RD-DQ-G\", \"global performance input chemical demand of oxygen\"),\n", + " (\"RD-SSG\", \"global performance input suspended solids\"),\n", + " (\"RD-SED-G\", \"global performance input sediments\"),\n", + "])\n", + "\n", + "data = pd.read_csv(\"data/water-treatment.data\", names=columns.keys())\n", + "data = data.replace('?', np.nan)" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": { + "collapsed": false, + "scrolled": false + }, + "outputs": [], + "source": [ + "# Capture only the numeric columns in the data set. \n", + "numeric_columns = columns.keys()\n", + "numeric_columns.remove(\"DAY\")\n", + "data = data[numeric_columns].apply(pd.to_numeric)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## 2D Rank Features " + ] + }, + { + "cell_type": "code", + "execution_count": 40, + "metadata": { + "collapsed": false, + "scrolled": false + }, + "outputs": [], + "source": [ + "def apply_column_pairs(func):\n", + " \"\"\"\n", + " Applies a function to a pair of columns and returns a new \n", + " dataframe that contains the result of the function as a matrix\n", + " of each pair of columns. \n", + " \"\"\"\n", + " \n", + " def inner(df):\n", + " cols = pd.DataFrame([\n", + " [\n", + " func(df[acol], df[bcol]) for bcol in df.columns\n", + " ] for acol in df.columns\n", + " ])\n", + "\n", + " cols.columns = df.columns\n", + " cols.index = df.columns \n", + " return cols\n", + "\n", + " return inner \n", + "\n", + "\n", + "@apply_column_pairs\n", + "def least_square_error(cola, colb):\n", + " \"\"\"\n", + " Computes the Root Mean Squared Error of a linear regression \n", + " between two columns of data. \n", + " \"\"\"\n", + " x = cola.fillna(np.nanmean(cola))\n", + " y = colb.fillna(np.nanmean(colb))\n", + " \n", + " m, b = np.polyfit(x, y, 1)\n", + " yh = (x * m) + b \n", + " return ((y-yh) ** 2).mean()" + ] + }, + { + "cell_type": "code", + "execution_count": 41, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAk8AAAIWCAYAAABORmlRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n", + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "labeled_metrics = {\n", + " 'Pearson': 'pearson', \n", + " 'Kendall Tao': 'kendall', \n", + " 'Spearman': 'spearman', \n", + " 'Pairwise Covariance': 'covariance',\n", + " 'Least Squares Error': 'lse', \n", + "}\n", + "\n", + "@interact(metric=labeled_metrics, data=fixed(data))\n", + "def rank2d(data, metric='pearson'):\n", + " \"\"\"\n", + " Creates a visualization of pairwise ranking by column in the data. \n", + " \"\"\"\n", + " \n", + " # The different rank by 2d metrics. \n", + " metrics = {\n", + " \"pearson\": lambda df: df.corr('pearson'), \n", + " \"kendall\": lambda df: df.corr('kendall'), \n", + " \"spearman\": lambda df: df.corr('spearman'), \n", + " \"covariance\": lambda df: df.cov(), \n", + " \"lse\": least_square_error,\n", + " }\n", + " \n", + " # Quick check to make sure a valid metric is passed in. \n", + " if metric not in metrics:\n", + " raise ValueError(\n", + " \"'{}' not a valid metric, specify one of {}\".format(\n", + " metric, \", \".join(metrics.keys())\n", + " )\n", + " )\n", + " \n", + " \n", + " # Compute the correlation matrix\n", + " corr = metrics[metric](data)\n", + "\n", + " # Generate a mask for the upper triangle\n", + " mask = np.zeros_like(corr, dtype=np.bool)\n", + " mask[np.triu_indices_from(mask)] = True\n", + "\n", + " # Set up the matplotlib figure\n", + " f, ax = plt.subplots(figsize=(11, 9))\n", + " ax.set_title(\"{} metric across {} features\".format(metric.title(), len(data.columns)))\n", + " \n", + " # Draw the heatmap with the mask and correct aspect ratio\n", + " sns.heatmap(corr, mask=mask, vmax=.3,\n", + " square=True, xticklabels=5, yticklabels=5,\n", + " linewidths=.5, cbar_kws={\"shrink\": .5}, ax=ax)" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 2", + "language": "python", + "name": "python2" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 2 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython2", + "version": "2.7.11" + }, + "widgets": { + "state": { + "2b2e97ac5fb04360a14b241b102d297c": { + "views": [ + { + "cell_index": 7 + } + ] + } + }, + "version": "1.2.0" + } + }, + "nbformat": 4, + "nbformat_minor": 0 +} diff --git a/images/yellowbrickroad.jpg b/images/yellowbrickroad.jpg new file mode 100644 index 000000000..c7106d30b Binary files /dev/null and b/images/yellowbrickroad.jpg differ diff --git a/mkdocs.yml b/mkdocs.yml deleted file mode 100644 index c8e810a23..000000000 --- a/mkdocs.yml +++ /dev/null @@ -1,10 +0,0 @@ -site_name: Yellowbrick -theme: readthedocs -repo_name: GitHub -repo_url: https://github.com/DistrictDataLabs/yellowbrick -site_description: A suite of visual analysis and diagnostic tools for machine learning. -site_author: District Data Labs -copyright: "© 2016 District Data Labs, All Rights Reserved" - -pages: - - "Introduction": index.md diff --git a/requirements.txt b/requirements.txt index c3504e684..46a4e65b8 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,10 +1,11 @@ # Dependencies matplotlib==1.5.1 +scipy==0.17.1 scikit-learn==0.17.1 numpy==1.11.0 +cycler==0.10.0 ## Utilities -#cycler==0.10.0 #pyparsing==2.1.4 #pytz==2016.4 #python-dateutil==2.5.3 @@ -13,13 +14,17 @@ numpy==1.11.0 ## Testing Requirements (uncomment for development) #nose==1.3.7 -#coverage==4.0.3 +#coverage==4.1 ## Build Requirements (uncomment for deployment) #wheel==0.29.0 ## Pip stuff (ignore) -#Python==2.7.11 +#Python==3.5.1 #pip==8.1.2 -#setuptools==21.0.0 +#setuptools==22.0.5 #wsgiref==0.1.2 + +## Documentation (uncomment to build documentation) +#Sphinx==1.4.4 +#sphinx-rtd-theme==0.1.9 \ No newline at end of file diff --git a/setup.py b/setup.py index fb48d830a..a2f42b026 100755 --- a/setup.py +++ b/setup.py @@ -8,7 +8,7 @@ # Copyright (C) 2016 District Data Labs # For license information, see LICENSE.txt and NOTICE.md # -# ID: setup.py [] benjamin@bengfort.com $ +# ID: setup.py [c4f3ba7] benjamin@bengfort.com $ """ Setup script for installing yellowbrick. diff --git a/tests/__init__.py b/tests/__init__.py index 7127a6a2a..d6493e068 100644 --- a/tests/__init__.py +++ b/tests/__init__.py @@ -7,7 +7,7 @@ # Copyright (C) 2016 District Data Labs # For license information, see LICENSE.txt # -# ID: tests.py [] benjamin@bengfort.com $ +# ID: __init__.py [0c5ba04] benjamin@bengfort.com $ """ Testing package for the yellowbrick visualization library. @@ -24,7 +24,7 @@ ## Test Constants ########################################################################## -EXPECTED_VERSION = "0.1" +EXPECTED_VERSION = "0.2" ########################################################################## diff --git a/tests/test_bestfit.py b/tests/test_bestfit.py new file mode 100644 index 000000000..cf9c706d6 --- /dev/null +++ b/tests/test_bestfit.py @@ -0,0 +1,128 @@ +# tests.test_bestfit +# Tests for the bestfit module. +# +# Author: Benjamin Bengfort +# Created: Sun Jun 26 19:27:39 2016 -0400 +# +# Copyright (C) 2016 District Data Labs +# For license information, see LICENSE.txt +# +# ID: test_bestfit.py [56236f3] benjamin@bengfort.com $ + +""" +Tests for the bestfit module. +""" + +########################################################################## +## Imports +########################################################################## + +import unittest +import numpy as np +import matplotlib.pyplot as plt + +from yellowbrick.bestfit import * +from yellowbrick.anscombe import ANSCOMBE +from yellowbrick.exceptions import YellowbrickValueError +from sklearn.linear_model import LinearRegression +from sklearn.pipeline import Pipeline + + +########################################################################## +## Best fit tests +########################################################################## + +class BestFitTests(unittest.TestCase): + + def test_bad_estimator(self): + """ + Test that a bad estimator name raises a value error. + """ + fig, axe = plt.subplots() + X, y = ANSCOMBE[1] + + with self.assertRaises(YellowbrickValueError): + draw_best_fit(X, y, axe, 'pepper') + + def test_ensure_same_length(self): + """ + Ensure that vectors of different lengths raise + """ + fig, axe = plt.subplots() + X = np.array([1, 2, 3, 5, 8, 10, 2]) + y = np.array([1, 3, 6, 2]) + + with self.assertRaises(YellowbrickValueError): + draw_best_fit(X, y, axe, 'linear') + + with self.assertRaises(YellowbrickValueError): + draw_best_fit(X[:,np.newaxis], y, axe, 'linear') + + def test_draw_best_fit(self): + """ + Test that drawing a best fit line works. + """ + fig, axe = plt.subplots() + X, y = ANSCOMBE[0] + + self.assertEqual(axe, draw_best_fit(X, y, axe, 'linear')) + self.assertEqual(axe, draw_best_fit(X, y, axe, 'quadratic')) + + +########################################################################## +## Estimator tests +########################################################################## + +class EstimatorTests(unittest.TestCase): + """ + Test the estimator functions for best fit lines. + """ + + def test_linear(self): + """ + Test the linear best fit estimator + """ + X, y = ANSCOMBE[0] + X = np.array(X) + y = np.array(y) + X = X[:,np.newaxis] + + model = fit_linear(X, y) + self.assertIsNotNone(model) + self.assertIsInstance(model, LinearRegression) + + + def test_quadratic(self): + """ + Test the quadratic best fit estimator + """ + X, y = ANSCOMBE[1] + X = np.array(X) + y = np.array(y) + X = X[:,np.newaxis] + + model = fit_quadratic(X, y) + self.assertIsNotNone(model) + self.assertIsInstance(model, Pipeline) + + def test_select_best(self): + """ + Test the select best fit estimator + """ + X, y = ANSCOMBE[1] + X = np.array(X) + y = np.array(y) + X = X[:,np.newaxis] + + model = fit_select_best(X, y) + self.assertIsNotNone(model) + self.assertIsInstance(model, Pipeline) + + X, y = ANSCOMBE[3] + X = np.array(X) + y = np.array(y) + X = X[:,np.newaxis] + + model = fit_select_best(X, y) + self.assertIsNotNone(model) + self.assertIsInstance(model, LinearRegression) diff --git a/tests/test_regressor.py b/tests/test_regressor.py new file mode 100644 index 000000000..8026db6e8 --- /dev/null +++ b/tests/test_regressor.py @@ -0,0 +1,52 @@ +# tests.test_regressor +# Ensure that the regressor visualizations work. +# +# Author: Benjamin Bengfort +# Created: Fri Jun 03 14:20:02 2016 -0700 +# +# Copyright (C) 2016 District Data Labs +# For license information, see LICENSE.txt +# +# ID: test_regressor.py [be63645] benjamin@bengfort.com $ + +""" +Ensure that the regressor visualizations work. +""" + +########################################################################## +## Imports +########################################################################## + +import unittest + +from yellowbrick.regressor import * +from yellowbrick.utils import * + +from sklearn.ensemble import RandomForestRegressor +from sklearn.svm import SVR + +########################################################################## +## Prediction error test case +########################################################################## + +class PredictionErrorTests(unittest.TestCase): + + def test_init_pe_viz(self): + """ + Ensure that both a single model and multiple models can be rendered + """ + viz = PredictionError([RandomForestRegressor(), SVR()]) + self.assertEqual(len(viz.models), 2) + + viz = PredictionError(SVR()) + self.assertEqual(len(viz.models), 1) + + def test_init_pe_names(self): + """ + Ensure that model names are correctly extracted + """ + viz = PredictionError([RandomForestRegressor(), SVR()]) + self.assertEqual(viz.names, ["RandomForestRegressor", "SVR"]) + + viz = PredictionError(SVR()) + self.assertEqual(viz.names, ["SVR"]) diff --git a/tests/test_utils.py b/tests/test_utils.py new file mode 100644 index 000000000..1b041de86 --- /dev/null +++ b/tests/test_utils.py @@ -0,0 +1,97 @@ +# tests.test_utils +# Test the export module - to generate a corpus for machine learning. +# +# Author: Jason Keung +# Patrick O'Melveny +# Created: Thurs Jun 2 15:33:18 2016 -0500 +# +# For license information, see LICENSE.txt +# + + +""" +Test the utils module - to generate a corpus for machine learning. +""" + +########################################################################## +## Imports +########################################################################## + +from sklearn.decomposition import PCA +from sklearn.linear_model import LassoCV +from sklearn.linear_model import LinearRegression +from sklearn.neighbors import LSHForest +from sklearn.pipeline import Pipeline +import unittest + +from yellowbrick.utils import get_model_name, isestimator + + +class ModelNameTests(unittest.TestCase): + + def test_real_model(self): + """ + Test that model name works for sklearn estimators + """ + model1 = LassoCV() + model2 = LSHForest() + self.assertEqual(get_model_name(model1), 'LassoCV') + self.assertEqual(get_model_name(model2), 'LSHForest') + + def test_pipeline(self): + """ + Test that model name works for sklearn pipelines + """ + pipeline = Pipeline([('reduce_dim', PCA()), + ('linreg', LinearRegression())]) + self.assertEqual(get_model_name(pipeline), 'LinearRegression') + + def test_int_input(self): + """ + Assert a type error is raised when an int is passed to model name. + """ + self.assertRaises(TypeError, get_model_name, 1) + + def test_str_input(self): + """ + Assert a type error is raised when a str is passed to model name. + """ + self.assertRaises(TypeError, get_model_name, 'helloworld') + + def test_estimator_instance(self): + """ + Test that isestimator works for instances + """ + model = LinearRegression() + self.assertTrue(isestimator(model)) + + def test_pipeline_instance(self): + """ + Test that isestimator works for pipelines + """ + model = Pipeline([ + ('reduce_dim', PCA()), + ('linreg', LinearRegression()) + ]) + + self.assertTrue(isestimator(model)) + + def test_estimator_class(self): + """ + Test that isestimator works for classes + """ + self.assertTrue(LinearRegression) + + def test_collection_not_estimator(self): + """ + Make sure that a collection is not an estimator + """ + for cls in (list, dict, tuple, set): + self.assertFalse(isestimator(cls)) + + things = ['pepper', 'sauce', 'queen'] + self.assertFalse(isestimator(things)) + + +if __name__ == "__main__": + unittest.main() diff --git a/tests/test_yb_palettes.py b/tests/test_yb_palettes.py new file mode 100644 index 000000000..986b8d6d9 --- /dev/null +++ b/tests/test_yb_palettes.py @@ -0,0 +1,147 @@ +# tests.test_yb_palettes +# Test the export module - to generate a corpus for machine learning. +# +# Author: Patrick O'Melveny +# Created: Friaday Jun 3 +# +# For license information, see LICENSE.txt +# + + +########################################################################## +## Imports +########################################################################## +import unittest +import numpy as np +import matplotlib as mpl +import warnings + + +from yellowbrick import yb_palettes, yb_rcmod, color_utils + + +class TestColorPalettes(unittest.TestCase): + + def test_current_palette(self): + + pal = yb_palettes.color_palette(["red", "blue", "green"], 3) + yb_rcmod.set_palette(pal, 3) + self.assertEqual(pal, color_utils.get_color_cycle()) + yb_rcmod.set() + + def test_palette_context(self): + + default_pal = yb_palettes.color_palette() + context_pal = yb_palettes.color_palette("muted") + + with yb_palettes.color_palette(context_pal): + self.assertEqual(color_utils.get_color_cycle(), context_pal) + + self.assertEqual(color_utils.get_color_cycle(), default_pal) + + def test_big_palette_context(self): + + original_pal = yb_palettes.color_palette("accent", n_colors=8) + context_pal = yb_palettes.color_palette("bold", 10) + + yb_rcmod.set_palette(original_pal) + with yb_palettes.color_palette(context_pal, 10): + self.assertEqual(color_utils.get_color_cycle(), context_pal) + + self.assertEqual(color_utils.get_color_cycle(), original_pal) + + # Reset default + yb_rcmod.set() + + def test_yellowbrick_palettes(self): + + pals = ["accent", "dark", "paired", "pastel", "bold", "muted"] + for name in pals: + pal_out = yb_palettes.color_palette(name) + self.assertEqual(len(pal_out), 6 if name != 'paired' else 10) + + def test_seaborn_palettes(self): + + pals = ["sns_deep", "sns_muted", "sns_pastel", + "sns_bright", "sns_dark", "sns_colorblind"] + for name in pals: + pal_out = yb_palettes.color_palette(name) + self.assertEqual(len(pal_out), 6) + + def test_bad_palette_name(self): + + with self.assertRaises(ValueError): + yb_palettes.color_palette("IAmNotAPalette") + + def test_terrible_palette_name(self): + + with self.assertRaises(ValueError): + yb_palettes.color_palette("jet") + + def test_bad_palette_colors(self): + + pal = ["red", "blue", "iamnotacolor"] + with self.assertRaises(ValueError): + yb_palettes.color_palette(pal) + + def test_palette_is_list_of_tuples(self): + + pal_in = np.array(["red", "blue", "green"]) + pal_out = yb_palettes.color_palette(pal_in, 3) + + self.assertIsInstance(pal_out, list) + self.assertIsInstance(pal_out[0], tuple) + self.assertIsInstance(pal_out[0][0], float) + self.assertEqual(len(pal_out[0]), 3) + + def test_palette_cycles(self): + + accent = yb_palettes.color_palette("accent") + double_accent = yb_palettes.color_palette("accent", 12) + self.assertEqual(double_accent, accent + accent) + + """ + def test_cbrewer_qual(self): + + pal_short = yb_palettes.mpl_palette("Set1", 4) + pal_long = yb_palettes.mpl_palette("Set1", 6) + self.assertEqual(pal_short, pal_long[:4]) + + pal_full = palettes.mpl_palette("Set2", 8) + pal_long = palettes.mpl_palette("Set2", 10) + self.assertEqual(pal_full, pal_long[:8]) + """ + + def test_color_codes(self): + + yb_palettes.set_color_codes("accent") + colors = yb_palettes.color_palette("accent") + [".1"] + for code, color in zip("bgrmyck", colors): + rgb_want = mpl.colors.colorConverter.to_rgb(color) + rgb_got = mpl.colors.colorConverter.to_rgb(code) + self.assertEqual(rgb_want, rgb_got) + yb_palettes.set_color_codes("reset") + + def test_as_hex(self): + + pal = yb_palettes.color_palette("accent") + for rgb, hex in zip(pal, pal.as_hex()): + self.assertEqual(mpl.colors.rgb2hex(rgb), hex) + """ + def test_preserved_palette_length(self): + + pal_in = palettes.color_palette("Set1", 10) + pal_out = palettes.color_palette(pal_in) + nt.assert_equal(pal_in, pal_out) + """ + + def test_get_color_cycle(self): + with warnings.catch_warnings(): + warnings.simplefilter('ignore') + result = color_utils.get_color_cycle() + expected = mpl.rcParams['axes.color_cycle'] + self.assertEqual(result, expected) + + +if __name__ == "__main__": + unittest.main() diff --git a/tests/test_yb_rcmod.py b/tests/test_yb_rcmod.py new file mode 100644 index 000000000..95e39da50 --- /dev/null +++ b/tests/test_yb_rcmod.py @@ -0,0 +1,194 @@ +# tests.test_utils +# Test the export module - to generate a corpus for machine learning. +# +# Author: Patrick O'Melveny +# Created: Thurs Jun 3 +# +# For license information, see LICENSE.txt +# + + +########################################################################## +## Imports +########################################################################## +import unittest +import numpy.testing as npt +import numpy as np +import matplotlib as mpl +from distutils.version import LooseVersion + + +from yellowbrick import yb_rcmod + + +class RCParamTester(unittest.TestCase): + + def flatten_list(self, orig_list): + + iter_list = map(np.atleast_1d, orig_list) + flat_list = [item for sublist in iter_list for item in sublist] + return flat_list + + def assert_rc_params(self, params): + + for k, v in params.items(): + if k == "svg.embed_char_paths": + # This param causes test issues and is deprecated anyway + continue + elif isinstance(v, np.ndarray): + npt.assert_array_equal(mpl.rcParams[k], v) + else: + self.assertEqual((k, mpl.rcParams[k]), (k, v)) + + +class TestAxesStyle(RCParamTester): + + styles = ["white", "dark", "whitegrid", "darkgrid", "ticks"] + + def test_default_return(self): + + current = yb_rcmod.axes_style() + self.assert_rc_params(current) + + def test_key_usage(self): + + _style_keys = set(yb_rcmod._style_keys) + for style in self.styles: + self.assertTrue(not set(yb_rcmod.axes_style(style)) ^ _style_keys) + + def test_bad_style(self): + + with self.assertRaises(ValueError): + yb_rcmod.axes_style("i_am_not_a_style") + + def test_rc_override(self): + + rc = {"axes.facecolor": "blue", "foo.notaparam": "bar"} + out = yb_rcmod.axes_style("darkgrid", rc) + self.assertEqual(out["axes.facecolor"], "blue") + self.assertNotIn("foo.notaparam", out) + + def test_set_style(self): + + for style in self.styles: + + style_dict = yb_rcmod.axes_style(style) + yb_rcmod.set_style(style) + self.assert_rc_params(style_dict) + + def test_style_context_manager(self): + + yb_rcmod.set_style("darkgrid") + orig_params = yb_rcmod.axes_style() + context_params = yb_rcmod.axes_style("whitegrid") + + with yb_rcmod.axes_style("whitegrid"): + self.assert_rc_params(context_params) + self.assert_rc_params(orig_params) + + @yb_rcmod.axes_style("whitegrid") + def func(): + self.assert_rc_params(context_params) + func() + self.assert_rc_params(orig_params) + + def test_style_context_independence(self): + + self.assertTrue(set(yb_rcmod._style_keys) ^ set(yb_rcmod._context_keys)) + + def test_set_rc(self): + + yb_rcmod.set(rc={"lines.linewidth": 4}) + self.assertEqual(mpl.rcParams["lines.linewidth"], 4) + yb_rcmod.set() + + def test_reset_defaults(self): + + # Changes to the rc parameters make this test hard to manage + # on older versions of matplotlib, so we'll skip it + if LooseVersion(mpl.__version__) < LooseVersion("1.3"): + raise self.SkipTest + + yb_rcmod.reset_defaults() + self.assert_rc_params(mpl.rcParamsDefault) + yb_rcmod.set() + + def test_reset_orig(self): + + # Changes to the rc parameters make this test hard to manage + # on older versions of matplotlib, so we'll skip it + if LooseVersion(mpl.__version__) < LooseVersion("1.3"): + raise self.SkipTest + + yb_rcmod.reset_orig() + self.assert_rc_params(mpl.rcParamsOrig) + yb_rcmod.set() + + +class TestPlottingContext(RCParamTester): + + contexts = ["paper", "notebook", "talk", "poster"] + + def test_default_return(self): + + current = yb_rcmod.plotting_context() + self.assert_rc_params(current) + + def test_key_usage(self): + + _context_keys = set(yb_rcmod._context_keys) + for context in self.contexts: + missing = set(yb_rcmod.plotting_context(context)) ^ _context_keys + self.assertTrue(not missing) + + def test_bad_context(self): + + with self.assertRaises(ValueError): + yb_rcmod.plotting_context("i_am_not_a_context") + + def test_font_scale(self): + + notebook_ref = yb_rcmod.plotting_context("notebook") + notebook_big = yb_rcmod.plotting_context("notebook", 2) + + font_keys = ["axes.labelsize", "axes.titlesize", "legend.fontsize", + "xtick.labelsize", "ytick.labelsize", "font.size"] + + for k in font_keys: + self.assertEqual(notebook_ref[k] * 2, notebook_big[k]) + + def test_rc_override(self): + + key, val = "grid.linewidth", 5 + rc = {key: val, "foo": "bar"} + out = yb_rcmod.plotting_context("talk", rc=rc) + self.assertEqual(out[key], val) + self.assertNotIn("foo", out) + + def test_set_context(self): + + for context in self.contexts: + + context_dict = yb_rcmod.plotting_context(context) + yb_rcmod.set_context(context) + self.assert_rc_params(context_dict) + + def test_context_context_manager(self): + + yb_rcmod.set_context("notebook") + orig_params = yb_rcmod.plotting_context() + context_params = yb_rcmod.plotting_context("paper") + + with yb_rcmod.plotting_context("paper"): + self.assert_rc_params(context_params) + self.assert_rc_params(orig_params) + + @yb_rcmod.plotting_context("paper") + def func(): + self.assert_rc_params(context_params) + func() + self.assert_rc_params(orig_params) + + +if __name__ == "__main__": + unittest.main() diff --git a/yellowbrick/__init__.py b/yellowbrick/__init__.py index f7e1e28ff..49e7cb917 100644 --- a/yellowbrick/__init__.py +++ b/yellowbrick/__init__.py @@ -7,7 +7,7 @@ # Copyright (C) 2016 District Data Labs # For license information, see LICENSE.txt # -# ID: __init__.py [] benjamin@bengfort.com $ +# ID: __init__.py [0c5ba04] benjamin@bengfort.com $ """ A suite of visual analysis and diagnostic tools to facilitate feature @@ -17,10 +17,22 @@ ########################################################################## ## Imports ########################################################################## +# Capture the original matplotlib rcParams +import matplotlib as mpl +_orig_rc_params = mpl.rcParams.copy() from .version import get_version from .anscombe import anscombe -from .classifier import crplot, rocplot_compare +from .classifier import crplot, rocplot +from .regressor import peplot, residuals_plot +from .yb_rcmod import * +from .yb_palettes import * + +########################################################################## +## Set default aesthetics +########################################################################## + +set() ########################################################################## ## Package Version diff --git a/yellowbrick/anscombe.py b/yellowbrick/anscombe.py index 91915cabc..b2d87bb45 100644 --- a/yellowbrick/anscombe.py +++ b/yellowbrick/anscombe.py @@ -7,7 +7,7 @@ # Copyright (C) 2016 District Data Labs # For license information, see LICENSE.txt # -# ID: anscombe.py [] benjamin@bengfort.com $ +# ID: anscombe.py [0bfa366] benjamin@bengfort.com $ """ Plots Anscombe's Quartet as an illustration of the importance of visualization. @@ -17,10 +17,11 @@ ## Imports ########################################################################## - import numpy as np import matplotlib.pyplot as plt +from yellowbrick.bestfit import draw_best_fit + ########################################################################## ## Anscombe Data Arrays @@ -55,9 +56,19 @@ def anscombe(): x = arr[0] y = arr[1] + # Set the X and Y limits + ax.set_xlim(0, 15) + ax.set_ylim(0, 15) + + # Draw the points in the scatter plot ax.scatter(x, y, c='g') - m,b = np.polyfit(x, y, 1) - X = np.linspace(ax.get_xlim()[0], ax.get_xlim()[1], 100) - ax.plot(X, m*X+b, '-') + + # Draw the linear best fit line on the plot + draw_best_fit(x, y, ax) return (axa, axb, axc, axd) + + +if __name__ == '__main__': + anscombe() + plt.show() diff --git a/yellowbrick/base.py b/yellowbrick/base.py new file mode 100644 index 000000000..d121ed61c --- /dev/null +++ b/yellowbrick/base.py @@ -0,0 +1,116 @@ +# yellowbrick.base +# Abstract base classes and interface for Yellowbrick. +# +# Author: Benjamin Bengfort +# Created: Fri Jun 03 10:20:59 2016 -0700 +# +# Copyright (C) 2016 District Data Labs +# For license information, see LICENSE.txt +# +# ID: base.py [4a59c49] benjamin@bengfort.com $ + +""" +Abstract base classes and interface for Yellowbrick. +""" + +import matplotlib.pyplot as plt + +from .exceptions import YellowbrickTypeError +from .utils import get_model_name, isestimator +from sklearn.base import BaseEstimator, TransformerMixin +from sklearn.cross_validation import cross_val_predict as cvp + +########################################################################## +## Base class hierarhcy +########################################################################## + +class BaseVisualization(object): + """ + The root of the visual object hierarchy that defines how yellowbrick + creates, stores, and renders visual artifcats using matplotlib. + """ + + def render(self): + """ + Render is the primary entry point for producing the visualization. + """ + raise NotImplementedError( + "All visualizations must specify their own render methodology" + ) + + +class FeatureVisualization(BaseVisualization, BaseEstimator, TransformerMixin): + """ + A feature visualization class accepts as input a DataFrame or Numpy array + in order to investigate features individually or together. + + FeatureVisualization is itself a transformer so that it can be used in + a Scikit-Learn Pipeline to perform automatic visual analysis during build. + """ + + def fit(self, X, y=None, **kwargs): + pass + + def transform(self, X): + pass + + def render(self, data=None): + """ + A feature visualization renders data. + """ + raise NotImplementedError( + "Please specify how to render the feature visualization" + ) + + +class ModelVisualization(BaseVisualization, BaseEstimator): + """ + A model visualization class accepts as input a Scikit-Learn estimator(s) + and is itself an estimator (to be included in a Pipeline) in order to + visualize the efficacy of a particular fitted model. + """ + + def fit(self, X, y=None, **kwargs): + pass + + def predict(self, X): + pass + + def render(self, model=None): + """ + A model visualization renders a model + """ + raise NotImplementedError( + "Please specify how to render the model visualization" + ) + + +class MultiModelMixin(object): + """ + Does predict for each of the models and generates subplots. + """ + + def __init__(self, models, **kwargs): + # Ensure models is a collection, if it's a single estimator then we + # wrap it in a list so that the API doesn't break during render. + if isestimator(models): + models = [models] + + # Keep track of the models + self.models = models + self.names = kwargs.pop('names', list(map(get_model_name, models))) + + def generate_subplots(self): + """ + Generates the subplots for the number of given models. + """ + _, axes = plt.subplots(len(self.models), sharex=True, sharey=True) + return axes + + def predict(self, X, y): + """ + Returns a generator containing the predictions for each of the + internal models (using cross_val_predict and a CV=12). + """ + for model in self.models: + yield cvp(model, X, y, cv=12) diff --git a/yellowbrick/bestfit.py b/yellowbrick/bestfit.py new file mode 100644 index 000000000..bbc027bbf --- /dev/null +++ b/yellowbrick/bestfit.py @@ -0,0 +1,189 @@ +# yellowbrick.bestfit +# Uses Scikit-Learn to compute a best fit function, then draws it in the plot. +# +# Author: Benjamin Bengfort +# Created: Sun Jun 26 17:27:08 2016 -0400 +# +# Copyright (C) 2016 District Data Labs +# For license information, see LICENSE.txt +# +# ID: bestfit.py [56236f3] benjamin@bengfort.com $ + +""" +Uses Scikit-Learn to compute a best fit function, then draws it in the plot. +""" + +########################################################################## +## Imports +########################################################################## + +import numpy as np + +from sklearn import linear_model +from sklearn.preprocessing import PolynomialFeatures +from sklearn.pipeline import make_pipeline +from sklearn.metrics import mean_squared_error as mse + +from operator import itemgetter +from yellowbrick.exceptions import YellowbrickValueError + + +########################################################################## +## Module Constants +########################################################################## + +# Names of the various estimator functions +LINEAR = 'linear' +QUADRATIC = 'quadratic' +EXPONENTIAL = 'exponential' +LOG = 'log' +SELECT_BEST = 'select_best' + + +########################################################################## +## Draw Line of Best Fit +########################################################################## + +def draw_best_fit(X, y, ax, estimator='linear', **kwargs): + """ + Uses Scikit-Learn to fit a model to X and y then uses the resulting model + to predict the curve based on the X values. This curve is drawn to the ax + (matplotlib axis) which must be passed as the third variable. + + The estimator function can be one of the following: + + 'linear': Uses OLS to fit the regression + 'quadratic': Uses OLS with Polynomial order 2 + 'exponential': Not implemented yet + 'log': Not implemented yet + 'select_best': Selects the best fit via MSE + + The remaining keyword arguments are passed to ax.plot to define and + describe the line of best fit. + """ + + # Estimators are the types of best fit lines that can be drawn. + estimators = { + LINEAR: fit_linear, # Uses OLS to fit the regression + QUADRATIC: fit_quadratic, # Uses OLS with Polynomial order 2 + EXPONENTIAL: fit_exponential, # Not implemented yet + LOG: fit_log, # Not implemented yet + SELECT_BEST: fit_select_best, # Selects the best fit via MSE + } + + # Check to make sure that a correct estimator value was passed in. + if estimator not in estimators: + raise YellowbrickValueError( + "'{}' not a valid type of estimator; choose from {}".format( + estimator, ", ".join(estimators.keys()) + ) + ) + + # Then collect the estimator function from the mapping. + estimator = estimators[estimator] + + # Ensure that X and y are the same length + if len(X) != len(y): + raise YellowbrickValueError(( + "X and y must have same length:" + " X len {} doesn't match y len {}!" + ).format(len(X), len(y))) + + # Ensure that X and y are np.arrays + X = np.array(X) + y = np.array(y) + + # Verify that X is a two dimensional array for Scikit-Learn esitmators + # and that its dimensions are (n, 1) where n is the number of rows. + if X.ndim < 2: + X = X[:,np.newaxis] # Reshape X into the correct dimensions + + if X.ndim > 2: + raise YellowbrickValueError( + "X must be a (1,) or (n,1) dimensional array not {}".format(x.shape) + ) + + # Verify that y is a (n,) dimensional array + if y.ndim > 1: + raise YellowbrickValueError( + "y must be a (1,) dimensional array not {}".format(y.shape) + ) + + # Uses the estimator to fit the data and get the model back. + model = estimator(X, y) + + # Plot line of best fit onto the axes that were passed in. + # TODO: determin if xlim or X.min(), X.max() are better params + xr = np.linspace(*ax.get_xlim(), num=100) + ax.plot(xr, model.predict(xr[:,np.newaxis]), **kwargs) + return ax + + +########################################################################## +## Estimator Functions +########################################################################## + +def fit_select_best(X, y): + """ + Selects the best fit of the estimators already implemented by choosing the + model with the smallest mean square error metric for the trained values. + """ + models = [fit(X,y) for fit in [fit_linear, fit_quadratic]] + errors = map(lambda model: mse(y, model.predict(X)), models) + + return min(zip(models, errors), key=itemgetter(1))[0] + + +def fit_linear(X, y): + """ + Uses OLS to fit the regression. + """ + model = linear_model.LinearRegression() + model.fit(X, y) + return model + + +def fit_quadratic(X, y): + """ + Uses OLS with Polynomial order 2. + """ + model = make_pipeline( + PolynomialFeatures(2), linear_model.LinearRegression() + ) + model.fit(X, y) + return model + + +def fit_exponential(X, y): + """ + Fits an exponential curve to the data. + """ + raise NotImplementedError("Exponential best fit lines are not implemented") + + +def fit_log(X, y): + """ + Fit a logrithmic curve to the data. + """ + raise NotImplementedError("Logrithmic best fit lines are not implemented") + + + +if __name__ == '__main__': + import os + import pandas as pd + import matplotlib.pyplot as plt + + path = os.path.join(os.path.dirname(__file__), "..", "examples", "data", "concrete.xls") + if not os.path.exists(path): + raise Exception("Could not find path for testing") + + xkey = 'Fine Aggregate (component 7)(kg in a m^3 mixture)' + ykey = 'Coarse Aggregate (component 6)(kg in a m^3 mixture)' + data = pd.read_excel(path) + + fig, axe = plt.subplots() + axe.scatter(data[xkey], data[ykey]) + draw_best_fit(data[xkey], data[ykey], axe, 'select_best') + + plt.show() diff --git a/yellowbrick/classifier.py b/yellowbrick/classifier.py index 6f4353bea..9a45b3fcb 100644 --- a/yellowbrick/classifier.py +++ b/yellowbrick/classifier.py @@ -7,7 +7,7 @@ # Copyright (C) 2016 District Data Labs # For license information, see LICENSE.txt # -# ID: classifier.py [] benjamin@bengfort.com $ +# ID: classifier.py [5eee25b] benjamin@bengfort.com $ """ Visualizations related to evaluating Scikit-Learn classification models @@ -24,107 +24,166 @@ from sklearn.metrics import roc_curve, auc from sklearn.metrics import classification_report -from .color import ddlheatmap +from .color_utils import ddlheatmap +from .utils import get_model_name, isestimator +from .base import ModelVisualization, MultiModelMixin +########################################################################## +## Classification Visualization Base Object +########################################################################## + +class ClassifierVisualization(ModelVisualization): + pass + ########################################################################## ## Classification Report ########################################################################## -def crplot(model, y_true, y_pred, **kwargs): +class ClassifierReport(ClassifierVisualization): """ - Plots a classification report as a heatmap. (More to follow). + Classification report that shows the precision, recall, and F1 scores + for the model. Integrates numerical scores as well color-coded heatmap. """ - # Get classification report arguments - # TODO: Do a better job of guessing defaults from the model - cr_kwargs = { - 'labels': kwargs.pop('labels', None), - 'target_names': kwargs.pop('target_names', None), - 'sample_weight': kwargs.pop('sample_weight', None), - 'digits': kwargs.pop('digits', 2) - } - - # Generate the classification report - report = classification_report(y_true, y_pred, **cr_kwargs) - cmap = kwargs.pop('cmap', ddlheatmap) - title = kwargs.pop('title', '{} Classification Report'.format(model.__class__.__name__)) - - - # Parse classification report: move to it's own function - # TODO: make a bit more robust, or look for the sklearn util that doesn't stringify - lines = report.split('\n') - classes = [] - matrix = [] - - for line in lines[2:(len(lines)-3)]: - s = line.split() - classes.append(s[0]) - value = [float(x) for x in s[1: len(s) - 1]] - matrix.append(value) - - # Generate plots and figure - fig, ax = plt.subplots(1) - - for column in range(len(matrix)+1): - for row in range(len(classes)): - txt = matrix[row][column] - ax.text(column,row,matrix[row][column],va='center',ha='center') - - fig = plt.imshow(matrix, interpolation='nearest', cmap=cmap) - plt.title(title) - plt.colorbar() - x_tick_marks = np.arange(len(classes)+1) - y_tick_marks = np.arange(len(classes)) - plt.xticks(x_tick_marks, ['precision', 'recall', 'f1-score'], rotation=45) - plt.yticks(y_tick_marks, classes) - plt.ylabel('Classes') - plt.xlabel('Measures') - - return ax + def __init__(self, model, **kwargs): + self.model = model + self.cmap = kwargs.pop('cmap', ddlheatmap) + self.name = kwargs.pop('name', get_model_name(model)) + self.report = None -########################################################################## -## Receiver Operating Characteristics -########################################################################## + def parse_report(self): + """ + Custom classification_report parsing utility + """ -def rocplot_compare(models, y_true, y_pred, **kwargs): - """ - Plots a side by size comparison of the ROC plot with AUC metric embedded. - """ - if len(models) != len(y_true) and len(models) != len(y_pred): - raise ValueError( - "Pass in two models, two sets of target and predictions" - ) + if self.report is None: + raise ModelError("Call score() before generating the model for parsing.") + + # TODO: make a bit more robust, or look for the sklearn util that doesn't stringify + lines = self.report.split('\n') + classes = [] + matrix = [] + + for line in lines[2:(len(lines)-3)]: + s = line.split() + classes.append(s[0]) + value = [float(x) for x in s[1: len(s) - 1]] + matrix.append(value) + + return matrix, classes + + + def score(self, y_true, y_pred, **kwargs): + """ + Generates the Scikit-Learn classification_report + """ + # TODO: Do a better job of guessing defaults from the model + cr_kwargs = { + 'labels': kwargs.pop('labels', None), + 'target_names': kwargs.pop('target_names', None), + 'sample_weight': kwargs.pop('sample_weight', None), + 'digits': kwargs.pop('digits', 2) + } - # Set up split subplots for the curve comparison. - # TODO: ensure that the number of models is only 2 - fig, axes = plt.subplots(1, 2, sharey=True) + self.report = classification_report(y_true, y_pred, **cr_kwargs) - # Zip together each plot to generate them independently. - for model, y, yhat, ax in zip(models, y_true, y_pred, axes): - # Figure out the name of the model - if isinstance(model, Pipeline): - name = model.steps[-1][1].__class__.__name__ - else: - name = model.__class__.__name__ + def render(self): + """ + Renders the classification report across each axis. + """ + title = '{} Classification Report'.format(self.name) + matrix, classes = self.parse_report() - fpr, tpr, thresholds = roc_curve(y, yhat) - roc_auc = auc(fpr, tpr) + fig, ax = plt.subplots(1) - # Plot the ROC Curve with the specified AUC label. - ax.plot(fpr, tpr, c='#2B94E9', label='AUC = {:0.2f}'.format(roc_auc)) + for column in range(len(matrix)+1): + for row in range(len(classes)): + txt = matrix[row][column] + ax.text(column,row,matrix[row][column],va='center',ha='center') - # Plot the line of no discrimination to compare the curve to. - ax.plot([0,1],[0,1],'m--',c='#666666') + fig = plt.imshow(matrix, interpolation='nearest', cmap=self.cmap) + plt.title(title) + plt.colorbar() + x_tick_marks = np.arange(len(classes)+1) + y_tick_marks = np.arange(len(classes)) + plt.xticks(x_tick_marks, ['precision', 'recall', 'f1-score'], rotation=45) + plt.yticks(y_tick_marks, classes) + plt.ylabel('Classes') + plt.xlabel('Measures') - # Set the title and create the legend. - ax.set_title('ROC for {}'.format(name)) - ax.legend(loc='lower right') + return ax - # Refactor the limits of the plot - plt.xlim([0,1]) - plt.ylim([0,1.1]) - return axes +def crplot(model, y_true, y_pred, **kwargs): + """ + Plots a classification report as a heatmap. (More to follow). + """ + viz = ClassifierReport(model, **kwargs) + viz.score(y_true, y_pred, **kwargs) + + return viz.render() + + +########################################################################## +## Receiver Operating Characteristics +########################################################################## + +class ROCAUC(MultiModelMixin, ClassifierVisualization): + """ + Plot the ROC to visualize the tradeoff between the classifier's + sensitivity and specificity. + """ + def __init__(self, models, **kwargs): + """ + Pass in a collection of models to generate ROC curves. + """ + super(ROCAUC, self).__init__(models, **kwargs) + self.colors = { + 'roc': kwargs.pop('roc_color', '#2B94E9'), + 'diagonal': kwargs.pop('diagonal_color', '#666666'), + } + + def fit(self, X, y): + """ + Custom fit method + """ + self.models = list(map(lambda model: model.fit(X, y), self.models)) + + def render(self, X, y): + """ + Renders each ROC-AUC plot across each axis. + """ + for idx, axe in enumerate(self.generate_subplots()): + # Get the information for this axis + name = self.names[idx] + model = self.models[idx] + y_pred = model.predict(X) + fpr, tpr, thresholds = roc_curve(y, y_pred) + roc_auc = auc(fpr, tpr) + + axe.plot(fpr, tpr, c=self.colors['roc'], label='AUC = {:0.2f}'.format(roc_auc)) + + # Plot the line of no discrimination to compare the curve to. + axe.plot([0,1],[0,1],'m--',c=self.colors['diagonal']) + + axe.set_title('ROC for {}'.format(name)) + axe.legend(loc='lower right') + + plt.xlim([0,1]) + plt.ylim([0,1.1]) + + return axe + + +def rocplot(models, X, y, **kwargs): + """ + Take in the model, data and labels as input and generate a multi-plot of + the ROC plots with AUC metrics embedded. + """ + viz = ROCAUC(models, **kwargs) + viz.fit(X, y) + + return viz.render(X, y) diff --git a/yellowbrick/color.py b/yellowbrick/color.py deleted file mode 100644 index a1c5fcbd2..000000000 --- a/yellowbrick/color.py +++ /dev/null @@ -1,30 +0,0 @@ -# yellowbrick.color -# Defines color definitions and color maps specific to DDL and Yellowbrick. -# -# Author: Rebecca Bilbro -# Created: Wed May 18 12:41:35 2016 -0400 -# -# Copyright (C) 2016 District Data Labs -# For license information, see LICENSE.txt -# -# ID: color.py [] benjamin@bengfort.com $ - -""" -Defines color definitions and color maps specific to DDL and Yellowbrick. -""" - -########################################################################## -## Imports -########################################################################## - -from matplotlib import colors -from matplotlib.colors import ListedColormap - -########################################################################## -## Colors -########################################################################## - -ddl_heat = ['#DBDBDB','#DCD5CC','#DCCEBE','#DDC8AF','#DEC2A0','#DEBB91',\ - '#DFB583','#DFAE74','#E0A865','#E1A256','#E19B48','#E29539'] - -ddlheatmap = colors.ListedColormap(ddl_heat) diff --git a/yellowbrick/color_utils.py b/yellowbrick/color_utils.py new file mode 100644 index 000000000..8ad1959ba --- /dev/null +++ b/yellowbrick/color_utils.py @@ -0,0 +1,54 @@ +# yellowbrick.color_utils +# Defines functions related to colors and palettes. +# +# Author: Rebecca Bilbro +# Created: Wed May 18 12:41:35 2016 -0400 +# +# Copyright (C) 2016 District Data Labs +# For license information, see LICENSE.txt +# +# ID: color_utils.py [15f72bf] pvomelveny@gmail.com $ + +""" +Defines color definitions and color maps specific to DDL and Yellowbrick. +""" + +########################################################################## +## Imports +########################################################################## + +from __future__ import print_function, division +import matplotlib.colors as mplcol +import matplotlib as mpl + + +# Check to see if matplotlib is at least sorta up to date +from distutils.version import LooseVersion +mpl_ge_150 = LooseVersion(mpl.__version__) >= "1.5.0" + +# TODO: This block should probably be moved/removed soon +########################################################################## +## Compatability with old stuff for now +########################################################################## + +ddl_heat = ['#DBDBDB', '#DCD5CC', '#DCCEBE', '#DDC8AF', '#DEC2A0', '#DEBB91', + '#DFB583','#DFAE74', '#E0A865', '#E1A256', '#E19B48', '#E29539'] + +ddlheatmap = mplcol.ListedColormap(ddl_heat) + +########################################################################## +## Color Utils +########################################################################## + + +def get_color_cycle(): + if mpl_ge_150: + cyl = mpl.rcParams['axes.prop_cycle'] + # matplotlib 1.5 verifies that axes.prop_cycle *is* a cycler + # but no garuantee that there's a `color` key. + # so users could have a custom rcParmas w/ no color... + try: + return [x['color'] for x in cyl] + except KeyError: + pass # just return axes.color style below + return mpl.rcParams['axes.color_cycle'] diff --git a/yellowbrick/exceptions.py b/yellowbrick/exceptions.py new file mode 100644 index 000000000..726daa6b9 --- /dev/null +++ b/yellowbrick/exceptions.py @@ -0,0 +1,52 @@ +# yellowbrick.exceptions +# Exceptions hierarchy for the yellowbrick library +# +# Author: Benjamin Bengfort +# Created: Fri Jun 03 10:39:41 2016 -0700 +# +# Copyright (C) 2016 District Data Labs +# For license information, see LICENSE.txt +# +# ID: exceptions.py [cb75e0e] benjamin@bengfort.com $ + +""" +Exceptions hierarchy for the yellowbrick library +""" + +########################################################################## +## Exceptions Hierarchy +########################################################################## + +class YellowbrickError(Exception): + """ + The root exception for all yellowbrick related errors. + """ + pass + + +class VisualError(YellowbrickError): + """ + A problem when interacting with matplotlib or the display framework. + """ + pass + + +class ModelError(YellowbrickError): + """ + A problem when interacting with sklearn or the ML framework. + """ + pass + + +class YellowbrickTypeError(YellowbrickError, TypeError): + """ + There was an unexpected type or none for a property or input. + """ + pass + + +class YellowbrickValueError(YellowbrickError, ValueError): + """ + A bad value was passed into a function. + """ + pass diff --git a/yellowbrick/regressor.py b/yellowbrick/regressor.py new file mode 100644 index 000000000..f8c37f09e --- /dev/null +++ b/yellowbrick/regressor.py @@ -0,0 +1,159 @@ +# yellowbrick.regressor +# Visualizations related to evaluating Scikit-Learn regressor models +# +# Author: Benjamin Bengfort +# Created: Fri Jun 03 10:30:36 2016 -0700 +# +# Copyright (C) 2016 District Data Labs +# For license information, see LICENSE.txt +# +# ID: regressor.py [4a59c49] benjamin@bengfort.com $ + +""" +Visualizations related to evaluating Scikit-Learn regressor models +""" + +########################################################################## +## Imports +########################################################################## + +import matplotlib as mpl +import matplotlib.pyplot as plt + +from .bestfit import draw_best_fit +from .utils import get_model_name, isestimator +from .base import ModelVisualization, MultiModelMixin +from sklearn.cross_validation import train_test_split as tts + +########################################################################## +## Regression Visualization Base Object +########################################################################## + +class RegressorVisualization(ModelVisualization): + pass + + +########################################################################## +## Prediction Error Plots +########################################################################## + +class PredictionError(MultiModelMixin, RegressorVisualization): + + def __init__(self, models, **kwargs): + """ + Pass in a collection of models to generate prediction error graphs. + """ + super(PredictionError, self).__init__(models, **kwargs) + + self.colors = { + 'point': kwargs.pop('point_color', '#F2BE2C'), + 'line': kwargs.pop('line_color', '#2B94E9'), + } + + def render(self, X, y): + """ + Renders each of the scatter plots per matrix. + """ + for idx, (axe, y_pred) in enumerate(zip(self.generate_subplots(), self.predict(X, y))): + # Set the x and y limits + axe.set_xlim(y.min()-1, y.max()+1) + axe.set_ylim(y_pred.min()-1, y_pred.max()+1) + + # Plot the correct values + axe.scatter(y, y_pred, c=self.colors['point']) + + # Draw the linear best fit line on the regression + draw_best_fit(y, y_pred, axe, 'linear', ls='--', lw=2, c=self.colors['line']) + + # Set the title and the y-axis label + axe.set_title("Predicted vs. Actual Values for {}".format(self.names[idx])) + axe.set_ylabel('Predicted Value') + + # Finalize figure + plt.xlabel('Actual Value') + return axe # TODO: We shouldn't return the last axis + + +def peplot(models, X, y, **kwargs): + """ + Take in the model, data and labels as input and generate a multi-plot of + the prediction error for each model. + """ + viz = PredictionError(models, **kwargs) + return viz.render(X, y) + +########################################################################## +## Residuals Plots +########################################################################## + +class ResidualsPlot(MultiModelMixin, RegressorVisualization): + """ + Unlike PredictionError, this viz takes classes instead of model instances + we should revise the API to have FittedRegressorVisualization vs. etc. + + TODO: Fitted vs. Unfitted API. + """ + + def __init__(self, models, **kwargs): + """ + Pass in a collection of model classes to generate train/test residual + plots by fitting the models and ... someone finish this docstring. + """ + super(ResidualsPlot, self).__init__(models, **kwargs) + + # TODO: the names for the color arguments are _long_. + self.colors = { + 'train_point': kwargs.pop('train_point_color', '#2B94E9'), + 'test_point': kwargs.pop('test_point_color', '#94BA65'), + 'line': kwargs.pop('line_color', '#333333'), + } + + def fit(self, X, y): + """ + Fit all three models and also store the train/test splits. + + TODO: move to MultiModelMixin. + """ + # TODO: make test size a parameter and do better data storage on viz. + self.X_train, self.X_test, self.y_train, self.y_test = tts(X, y, test_size=0.2) + self.models = list(map(lambda model: model.fit(self.X_train, self.y_train), self.models)) + + def render(self): + """ + Renders each residual plot across each axis. + """ + + for idx, axe in enumerate(self.generate_subplots()): + # Get the information for this axis + model = self.models[idx] + name = self.names[idx] + + # TODO: less proceedural? + # Add the training residuals + y_train_pred = model.predict(self.X_train) + axe.scatter(y_train_pred, y_train_pred - self.y_train, c=self.colors['train_point'], s=40, alpha=0.5) + + # Add the test residuals + y_test_pred = model.predict(self.X_test) + axe.scatter(y_test_pred, y_test_pred - self.y_test, c=self.colors['test_point'], s=40) + + # Add the hline and other axis elements + # TODO: better parameters based on the plot or, normalize, then push -1 to 1 + axe.hlines(y=0, xmin=0, xmax=100) + axe.set_title(name) + axe.set_ylabel('Residuals') + + # Finalize the residuals plot + # TODO: adjust the x and y ranges in order to compare (or use normalize) + plt.xlabel("Predicted Value") + return axe # TODO: We shouldn't return the last axis + + +def residuals_plot(models, X, y, **kwargs): + """ + Take in the model, data and labels as input and generate a multi-plot of + the residuals for each. + """ + viz = ResidualsPlot(models, **kwargs) + viz.fit(X, y) + return viz.render() diff --git a/yellowbrick/utils.py b/yellowbrick/utils.py new file mode 100644 index 000000000..4b0ddd625 --- /dev/null +++ b/yellowbrick/utils.py @@ -0,0 +1,43 @@ +# utils +# +# Author: Jason Keung +# Patrick O'Melveny +# Created: Thurs Jun 2 15:33:18 2016 -0500 +# +# For license information, see LICENSE.txt + +""" +Utility functions for yellowbrick +""" + +########################################################################## +## Imports +########################################################################## + +from sklearn.pipeline import Pipeline +from sklearn.base import BaseEstimator + +########################################################################## +## Model detection utilities +########################################################################## + +def get_model_name(model): + """ + Detects the model name for a Scikit-Learn model or pipeline + """ + if not isinstance(model, BaseEstimator): + raise TypeError + else: + if isinstance(model, Pipeline): + return model.steps[-1][-1].__class__.__name__ + else: + return model.__class__.__name__ + +def isestimator(model): + """ + Determines if a model is an estimator using issubclass and isinstance. + """ + if type(model) == type: + return issubclass(model, BaseEstimator) + + return isinstance(model, BaseEstimator) diff --git a/yellowbrick/version.py b/yellowbrick/version.py index df1994d36..cfcb9a613 100644 --- a/yellowbrick/version.py +++ b/yellowbrick/version.py @@ -7,7 +7,7 @@ # Copyright (C) 2016 District Data Labs # For license information, see LICENSE.txt # -# ID: version.py [] benjamin@bengfort.com $ +# ID: version.py [0c5ba04] benjamin@bengfort.com $ """ Maintains version and package information for deployment. @@ -19,7 +19,7 @@ __version_info__ = { 'major': 0, - 'minor': 1, + 'minor': 2, 'micro': 0, 'releaselevel': 'final', 'serial': 0, diff --git a/yellowbrick/yb_palettes.py b/yellowbrick/yb_palettes.py new file mode 100644 index 000000000..d2e272fba --- /dev/null +++ b/yellowbrick/yb_palettes.py @@ -0,0 +1,198 @@ +# yellowbrick.yb_palettes +# Defines color definitions and color maps specific to DDL and Yellowbrick. +# +# Original based on Seaborn's rcmod.py: +# +# For license information, see LICENSE.txt +# +# TODO: Clean up docs so they don't reference Seaborn things we don't have + +"""Functions that alter the matplotlib rc dictionary on the fly.""" + +########################################################################## +## Imports +########################################################################## +from __future__ import division +from itertools import cycle + +import matplotlib as mpl +from six import string_types +from six.moves import range + +from .color_utils import get_color_cycle + + +########################################################################## +## Exports +########################################################################## +__all__ = ["color_palette", "set_color_codes"] + +########################################################################## +## Default Yellowbrick Palettes (and Default Seaborn, just cause) +########################################################################## +# Taken from Colorbrewer, qualitative color schemes +YELLOWBRICK_PALETTES = dict( + accent=['#7fc97f', '#beaed4', '#fdc086', + '#ffff99', '#386cb0', '#f0027f'], + dark=['#1b9e77', '#d95f02', '#7570b3', + '#e7298a', '#66a61e', '#e6ab02'], + paired=['#a6cee3', '#1f78b4', '#b2df8a', '#33a02c', + '#fb9a99', '#e31a1c', '#fdbf6f', '#ff7f00', + '#cab2d6', '#6a3d9a'], + pastel=['#b3e2cd', '#fdcdac', '#cbd5e8', + '#f4cae4', '#e6f5c9', '#fff2ae'], + bold=['#e41a1c', '#377eb8', '#4daf4a', + '#984ea3', '#ff7f00', '#ffff33'], + muted=['#8dd3c7', '#ffffb3', '#bebada', + '#fb8072', '#80b1d3', '#fdb462'] +) + +SEABORN_PALETTES = dict( + sns_deep=["#4C72B0", "#55A868", "#C44E52", + "#8172B2", "#CCB974", "#64B5CD"], + sns_muted=["#4878CF", "#6ACC65", "#D65F5F", + "#B47CC7", "#C4AD66", "#77BEDB"], + sns_pastel=["#92C6FF", "#97F0AA", "#FF9F9A", + "#D0BBFF", "#FFFEA3", "#B0E0E6"], + sns_bright=["#003FFF", "#03ED3A", "#E8000B", + "#8A2BE2", "#FFC400", "#00D7FF"], + sns_dark=["#001C7F", "#017517", "#8C0900", + "#7600A1", "#B8860B", "#006374"], + sns_colorblind=["#0072B2", "#009E73", "#D55E00", + "#CC79A7", "#F0E442", "#56B4E9"] + ) +########################################################################## +## Palette Functions +########################################################################## +class _ColorPalette(list): + """Set the color palette in a with statement, otherwise be a list.""" + def __enter__(self): + """Open the context.""" + from .yb_rcmod import set_palette + self._orig_palette = color_palette() + set_palette(self) + return self + + def __exit__(self, *args): + """Close the context.""" + from .yb_rcmod import set_palette + set_palette(self._orig_palette) + + def as_hex(self): + """Return a color palette with hex codes instead of RGB values.""" + hex = [mpl.colors.rgb2hex(rgb) for rgb in self] + return _ColorPalette(hex) + + +def color_palette(palette=None, n_colors=None, desat=None): + """Return a list of colors defining a color palette. + Calling this function with ``palette=None`` will return the current + matplotlib color cycle. + This function can also be used in a ``with`` statement to temporarily + set the color cycle for a plot or set of plots. + + :param palette: + Name of palette or None to return current palette. If a sequence, input + colors are used but possibly cycled and desaturated. + + Available seaborn palette names from :py:mod:`seaborn.palettes` are: + + .. hlist:: + :columns: 3 + + * :py:const:`deep` + * :py:const:`dark` + * :py:const:`paired` + * :py:const:`pastel` + * :py:const:`bold` + * :py:const:`muted` + * :py:const:`sns_deep` + * :py:const:`sns_muted` + * :py:const:`sns_bright` + * :py:const:`sns_pastel` + * :py:const:`sns_dark` + * :py:const:`sns_colorblind` + + :type palette: None or str or sequence + :param n_colors: + Number of colors in the palette. If ``None``, the default will depend + on how ``palette`` is specified. Named palettes default to 6 colors + (except paired, which has 10), + but grabbing the current palette or passing in a list of colors will + not change the number of colors unless this is specified. Asking for + more colors than exist in the palette will cause it to cycle. + :type n_colors: int or None + :param desat: + :type desat: + + :rtype: list(tuple) + :return: list of RGB tuples. + Color palette. Behaves like a list, but can be used as a context + manager and possesses an :py:meth:`as_hex` method to convert to hex color + codes. + + .. seealso:: + + :func:`.set_palette` + Set the default color cycle for all plots. + :func:`.set_color_codes` + Reassign color codes like ``"b"``, ``"g"``, etc. to + colors from one of the yellowbrick palettes. + """ + if palette is None: + palette = get_color_cycle() + if n_colors is None: + n_colors = len(palette) + + elif not isinstance(palette, string_types): + palette = palette + if n_colors is None: + n_colors = len(palette) + else: + if palette.lower() == "jet": + raise ValueError("No.") + elif palette in YELLOWBRICK_PALETTES: + palette = YELLOWBRICK_PALETTES[palette] + elif palette in SEABORN_PALETTES: + palette = SEABORN_PALETTES[palette] + else: + raise ValueError("%s is not a valid palette " + "name in yellowbrick" % palette) + if n_colors is None: + n_colors = len(palette) + + # Always return as many colors as we asked for + pal_cycle = cycle(palette) + palette = [next(pal_cycle) for _ in range(n_colors)] + + # Always return in r, g, b tuple format + try: + palette = map(mpl.colors.colorConverter.to_rgb, palette) + palette = _ColorPalette(palette) + except ValueError: + raise ValueError("Could not generate a palette for %s" % str(palette)) + + return palette + +def set_color_codes(palette="accent"): + """Change how matplotlib color shorthands are interpreted. + Calling this will change how shorthand codes like "b" or "g" + are interpreted by matplotlib in subsequent plots. + Parameters + ---------- + palette : {accent, dark, paired, pastel, bold, muted} + Named yellowbrick palette to use as the source of colors. + See Also + -------- + set_palette : Color codes can also be set through the function that + sets the matplotlib color cycle. + """ + if palette == "reset": + colors = [(0., 0., 1.), (0., .5, 0.), (1., 0., 0.), (.75, .75, 0.), + (.75, .75, 0.), (0., .75, .75), (0., 0., 0.)] + else: + colors = YELLOWBRICK_PALETTES[palette] + [(.1, .1, .1)] + for code, color in zip("bgrmyck", colors): + rgb = mpl.colors.colorConverter.to_rgb(color) + mpl.colors.colorConverter.colors[code] = rgb + mpl.colors.colorConverter.cache[code] = rgb diff --git a/yellowbrick/yb_rcmod.py b/yellowbrick/yb_rcmod.py new file mode 100644 index 000000000..a0863d5ea --- /dev/null +++ b/yellowbrick/yb_rcmod.py @@ -0,0 +1,494 @@ +# yellowbrick.yb_rcmod +# Defines color definitions and color maps specific to DDL and Yellowbrick. +# +# Original based on Seaborn's rcmod.py: +# +# For license information, see LICENSE.txt +# +# TODO: Clean up docs so they don't reference Seaborn things we don't have + +"""Functions that alter the matplotlib rc dictionary on the fly.""" + +########################################################################## +## Imports +########################################################################## +import functools + +import numpy as np +import matplotlib as mpl + +from six import string_types + +# Check to see if we have a slightly modern version of mpl +from distutils.version import LooseVersion +mpl_ge_150 = LooseVersion(mpl.__version__) >= '1.5.0' + +from . import yb_palettes, _orig_rc_params + + +########################################################################## +## Exports +########################################################################## +__all__ = ["set", "reset_defaults", "reset_orig", + "axes_style", "set_style", "plotting_context", "set_context", + "set_palette"] + + +########################################################################## +## Keys +########################################################################## +_style_keys = ( + + "axes.facecolor", + "axes.edgecolor", + "axes.grid", + "axes.axisbelow", + "axes.linewidth", + "axes.labelcolor", + + "figure.facecolor", + + "grid.color", + "grid.linestyle", + + "text.color", + + "xtick.color", + "ytick.color", + "xtick.direction", + "ytick.direction", + "xtick.major.size", + "ytick.major.size", + "xtick.minor.size", + "ytick.minor.size", + + "legend.frameon", + "legend.numpoints", + "legend.scatterpoints", + + "lines.solid_capstyle", + + "image.cmap", + "font.family", + "font.sans-serif", + ) + +_context_keys = ( + "figure.figsize", + + "font.size", + "axes.labelsize", + "axes.titlesize", + "xtick.labelsize", + "ytick.labelsize", + "legend.fontsize", + + "grid.linewidth", + "lines.linewidth", + "patch.linewidth", + "lines.markersize", + "lines.markeredgewidth", + + "xtick.major.width", + "ytick.major.width", + "xtick.minor.width", + "ytick.minor.width", + + "xtick.major.pad", + "ytick.major.pad" + ) + + +def set(context="notebook", style="darkgrid", palette="accent", + font="sans-serif", font_scale=1, color_codes=False, rc=None): + """Set aesthetic parameters in one step. + Each set of parameters can be set directly or temporarily, see the + referenced functions below for more information. + Parameters + ---------- + context : string or dict + Plotting context parameters, see :func:`plotting_context` + style : string or dict + Axes style parameters, see :func:`axes_style` + palette : string or sequence + Color palette, see :func:`color_palette` + font : string + Font family, see matplotlib font manager. + font_scale : float, optional + Separate scaling factor to independently scale the size of the + font elements. + color_codes : bool + If ``True`` and ``palette`` is a yellowbrick palette, remap the shorthand + color codes (e.g. "b", "g", "r", etc.) to the colors from this palette. + rc : dict or None + Dictionary of rc parameter mappings to override the above. + """ + set_context(context, font_scale) + set_style(style, rc={"font.family": font}) + set_palette(palette, color_codes=color_codes) + if rc is not None: + mpl.rcParams.update(rc) + + +def reset_defaults(): + """Restore all RC params to default settings.""" + mpl.rcParams.update(mpl.rcParamsDefault) + + +def reset_orig(): + """Restore all RC params to original settings (respects custom rc).""" + mpl.rcParams.update(_orig_rc_params) + + +########################################################################## +## Axes Styles +########################################################################## + +def axes_style(style=None, rc=None): + """Return a parameter dict for the aesthetic style of the plots. + This affects things like the color of the axes, whether a grid is + enabled by default, and other aesthetic elements. + This function returns an object that can be used in a ``with`` statement + to temporarily change the style parameters. + Parameters + ---------- + style : dict, None, or one of {darkgrid, whitegrid, dark, white, ticks} + A dictionary of parameters or the name of a preconfigured set. + rc : dict, optional + Parameter mappings to override the values in the preset seaborn + style dictionaries. This only updates parameters that are + considered part of the style definition. + Examples + -------- + >>> st = axes_style("whitegrid") + >>> set_style("ticks", {"xtick.major.size": 8, "ytick.major.size": 8}) + >>> import matplotlib.pyplot as plt + >>> with axes_style("white"): + ... f, ax = plt.subplots() + ... ax.plot(x, y) # doctest: +SKIP + See Also + -------- + set_style : set the matplotlib parameters for a seaborn theme + plotting_context : return a parameter dict to to scale plot elements + color_palette : define the color palette for a plot + """ + if style is None: + style_dict = {k: mpl.rcParams[k] for k in _style_keys} + + elif isinstance(style, dict): + style_dict = style + + else: + styles = ["white", "dark", "whitegrid", "darkgrid", "ticks"] + if style not in styles: + raise ValueError("style must be one of %s" % ", ".join(styles)) + + # Define colors here + dark_gray = ".15" + light_gray = ".8" + + # Common parameters + style_dict = { + "figure.facecolor": "white", + "text.color": dark_gray, + "axes.labelcolor": dark_gray, + "legend.frameon": False, + "legend.numpoints": 1, + "legend.scatterpoints": 1, + "xtick.direction": "out", + "ytick.direction": "out", + "xtick.color": dark_gray, + "ytick.color": dark_gray, + "axes.axisbelow": True, + "image.cmap": "Greys", + "font.family": ["sans-serif"], + "font.sans-serif": ["Arial", "Liberation Sans", + "Bitstream Vera Sans", "sans-serif"], + "grid.linestyle": "-", + "lines.solid_capstyle": "round", + } + + # Set grid on or off + if "grid" in style: + style_dict.update({ + "axes.grid": True, + }) + else: + style_dict.update({ + "axes.grid": False, + }) + + # Set the color of the background, spines, and grids + if style.startswith("dark"): + style_dict.update({ + "axes.facecolor": "#EAEAF2", + "axes.edgecolor": "white", + "axes.linewidth": 0, + "grid.color": "white", + }) + + elif style == "whitegrid": + style_dict.update({ + "axes.facecolor": "white", + "axes.edgecolor": light_gray, + "axes.linewidth": 1, + "grid.color": light_gray, + }) + + elif style in ["white", "ticks"]: + style_dict.update({ + "axes.facecolor": "white", + "axes.edgecolor": dark_gray, + "axes.linewidth": 1.25, + "grid.color": light_gray, + }) + + # Show or hide the axes ticks + if style == "ticks": + style_dict.update({ + "xtick.major.size": 6, + "ytick.major.size": 6, + "xtick.minor.size": 3, + "ytick.minor.size": 3, + }) + else: + style_dict.update({ + "xtick.major.size": 0, + "ytick.major.size": 0, + "xtick.minor.size": 0, + "ytick.minor.size": 0, + }) + + # Override these settings with the provided rc dictionary + if rc is not None: + rc = {k: v for k, v in rc.items() if k in _style_keys} + style_dict.update(rc) + + # Wrap in an _AxesStyle object so this can be used in a with statement + style_object = _AxesStyle(style_dict) + + return style_object + + +def set_style(style=None, rc=None): + """Set the aesthetic style of the plots. + This affects things like the color of the axes, whether a grid is + enabled by default, and other aesthetic elements. + Parameters + ---------- + style : dict, None, or one of {darkgrid, whitegrid, dark, white, ticks} + A dictionary of parameters or the name of a preconfigured set. + rc : dict, optional + Parameter mappings to override the values in the preset seaborn + style dictionaries. This only updates parameters that are + considered part of the style definition. + Examples + -------- + >>> set_style("whitegrid") + >>> set_style("ticks", {"xtick.major.size": 8, "ytick.major.size": 8}) + See Also + -------- + axes_style : return a dict of parameters or use in a ``with`` statement + to temporarily set the style. + set_context : set parameters to scale plot elements + set_palette : set the default color palette for figures + """ + style_object = axes_style(style, rc) + mpl.rcParams.update(style_object) + + +########################################################################## +## Context +########################################################################## +def plotting_context(context=None, font_scale=1, rc=None): + """Return a parameter dict to scale elements of the figure. + This affects things like the size of the labels, lines, and other + elements of the plot, but not the overall style. The base context + is "notebook", and the other contexts are "paper", "talk", and "poster", + which are version of the notebook parameters scaled by .8, 1.3, and 1.6, + respectively. + This function returns an object that can be used in a ``with`` statement + to temporarily change the context parameters. + Parameters + ---------- + context : dict, None, or one of {paper, notebook, talk, poster} + A dictionary of parameters or the name of a preconfigured set. + font_scale : float, optional + Separate scaling factor to independently scale the size of the + font elements. + rc : dict, optional + Parameter mappings to override the values in the preset seaborn + context dictionaries. This only updates parameters that are + considered part of the context definition. + Examples + -------- + >>> c = plotting_context("poster") + >>> c = plotting_context("notebook", font_scale=1.5) + >>> c = plotting_context("talk", rc={"lines.linewidth": 2}) + >>> import matplotlib.pyplot as plt + >>> with plotting_context("paper"): + ... f, ax = plt.subplots() + ... ax.plot(x, y) # doctest: +SKIP + See Also + -------- + set_context : set the matplotlib parameters to scale plot elements + axes_style : return a dict of parameters defining a figure style + color_palette : define the color palette for a plot + """ + if context is None: + context_dict = {k: mpl.rcParams[k] for k in _context_keys} + + elif isinstance(context, dict): + context_dict = context + + else: + + contexts = ["paper", "notebook", "talk", "poster"] + if context not in contexts: + raise ValueError("context must be in %s" % ", ".join(contexts)) + + # Set up dictionary of default parameters + base_context = { + + "figure.figsize": np.array([8, 5.5]), + "font.size": 12, + "axes.labelsize": 11, + "axes.titlesize": 12, + "xtick.labelsize": 10, + "ytick.labelsize": 10, + "legend.fontsize": 10, + + "grid.linewidth": 1, + "lines.linewidth": 1.75, + "patch.linewidth": .3, + "lines.markersize": 7, + "lines.markeredgewidth": 0, + + "xtick.major.width": 1, + "ytick.major.width": 1, + "xtick.minor.width": .5, + "ytick.minor.width": .5, + + "xtick.major.pad": 7, + "ytick.major.pad": 7, + } + + # Scale all the parameters by the same factor depending on the context + scaling = dict(paper=.8, notebook=1, talk=1.3, poster=1.6)[context] + context_dict = {k: v * scaling for k, v in base_context.items()} + + # Now independently scale the fonts + font_keys = ["axes.labelsize", "axes.titlesize", "legend.fontsize", + "xtick.labelsize", "ytick.labelsize", "font.size"] + font_dict = {k: context_dict[k] * font_scale for k in font_keys} + context_dict.update(font_dict) + + # Implement hack workaround for matplotlib bug + # See https://github.com/mwaskom/seaborn/issues/344 + # There is a bug in matplotlib 1.4.2 that makes points invisible when + # they don't have an edgewidth. It will supposedly be fixed in 1.4.3. + if mpl.__version__ == "1.4.2": + context_dict["lines.markeredgewidth"] = 0.01 + + # Override these settings with the provided rc dictionary + if rc is not None: + rc = {k: v for k, v in rc.items() if k in _context_keys} + context_dict.update(rc) + + # Wrap in a _PlottingContext object so this can be used in a with statement + context_object = _PlottingContext(context_dict) + + return context_object + + +def set_context(context=None, font_scale=1, rc=None): + """Set the plotting context parameters. + This affects things like the size of the labels, lines, and other + elements of the plot, but not the overall style. The base context + is "notebook", and the other contexts are "paper", "talk", and "poster", + which are version of the notebook parameters scaled by .8, 1.3, and 1.6, + respectively. + Parameters + ---------- + context : dict, None, or one of {paper, notebook, talk, poster} + A dictionary of parameters or the name of a preconfigured set. + font_scale : float, optional + Separate scaling factor to independently scale the size of the + font elements. + rc : dict, optional + Parameter mappings to override the values in the preset seaborn + context dictionaries. This only updates parameters that are + considered part of the context definition. + Examples + -------- + >>> set_context("paper") + >>> set_context("talk", font_scale=1.4) + >>> set_context("talk", rc={"lines.linewidth": 2}) + See Also + -------- + plotting_context : return a dictionary of rc parameters, or use in + a ``with`` statement to temporarily set the context. + set_style : set the default parameters for figure style + set_palette : set the default color palette for figures + """ + context_object = plotting_context(context, font_scale, rc) + mpl.rcParams.update(context_object) + + +class _RCAesthetics(dict): + def __enter__(self): + rc = mpl.rcParams + self._orig = {k: rc[k] for k in self._keys} + self._set(self) + + def __exit__(self, exc_type, exc_value, exc_tb): + self._set(self._orig) + + def __call__(self, func): + @functools.wraps(func) + def wrapper(*args, **kwargs): + with self: + return func(*args, **kwargs) + return wrapper + + +class _AxesStyle(_RCAesthetics): + """Light wrapper on a dict to set style temporarily.""" + _keys = _style_keys + _set = staticmethod(set_style) + + +class _PlottingContext(_RCAesthetics): + """Light wrapper on a dict to set context temporarily.""" + _keys = _context_keys + _set = staticmethod(set_context) + + +########################################################################## +## Colors/Palettes +########################################################################## +def set_palette(palette, n_colors=None, color_codes=False): + """Set the matplotlib color cycle using a seaborn palette. + Parameters + ---------- + palette : yellowbrick color palette | seaborn color palette (with sns_ prepended) + Palette definition. Should be something that :func:`color_palette` + can process. + n_colors : int + Number of colors in the cycle. The default number of colors will depend + on the format of ``palette``, see the :func:`color_palette` + documentation for more information. + color_codes : bool + If ``True`` and ``palette`` is a seaborn palette, remap the shorthand + color codes (e.g. "b", "g", "r", etc.) to the colors from this palette. + """ + colors = yb_palettes.color_palette(palette, n_colors) + if mpl_ge_150: + from cycler import cycler + cyl = cycler('color', colors) + mpl.rcParams['axes.prop_cycle'] = cyl + else: + mpl.rcParams["axes.color_cycle"] = list(colors) + mpl.rcParams["patch.facecolor"] = colors[0] + if color_codes: + yb_palettes.set_color_codes(palette)