Documentation on requirements, removed Numba.

I removed the now unused dependency on Numba.

Added description of requirements and testing in the README file.

README.md

@@ -45,6 +45,45 @@ from filterpy.memory import FadingMemoryFilter
 my_filter = KalmanFilter(3,4)
 ```
 
+Requirements
+--
+numpy and scipy
+Python 2 or 3
+matplotlib
+
+I haven't extensively tested backwards compatibility - I use the Anaconda
+distribution, and so I am on Python 3.4 and 2.7.5, along with whatever
+version of numpy, scipy, and matplotlib they provide. But I am using pretty
+basic Python - numpy.array, maybe a list comprehension in my tests.
+
+I import from __future__ to ensure the code works in Python 2 and 3.
+
+The matplotlib library is required because, *for now*, 'tests' are very
+visual. Meaning I generate some data, plot the data against the filtered
+results, and eyeball it. That is great for my personal development, and
+terrible as a foundation for regression testing. If you don't have
+matplotlib installed you won't be able to run the tests, but I'm not sure
+the tests will have a lot of meaning to you anyway.
+
+There is one import from the code from my book to plot ellipses. That
+dependency needs to be removed. This only affects the tests.
+
+
+Testing
+--
+
+All tests are written to work with py.test. Just type _py.text_ at the
+command line.
+
+As explained above, the tests are not robust. I'm still at the stage where
+visual plots are the best way to see how things are working. Apologies, but
+I think it is a sound choice for development. It is easy for a filter to
+perform within theoretical limits (which we can write a non-visual test for)
+yet be 'off' in some way. The code itself contains tests in the form of
+asserts and properties that ensure that arrays are of the proper dimension, 
+etc. 
+
+
 References
 --
 

leastsq/least_squares.py

@@ -16,8 +16,6 @@
 import numpy as np
 from math import sqrt
 
-from numba import jit, void
-
 
 class LeastSquaresFilter(object):
     """Implements a Least Squares recursive filter. Formulation is per

setup.py

@@ -5,7 +5,7 @@
 here = path.abspath(path.dirname(__file__))
 
 # Get the long description from the relevant file
-with open(path.join(here, 'DESCRIPTION.rst'), encoding='utf-8') as f:
+with open(path.join(here, 'README.md'), encoding='utf-8') as f:
     long_description = f.read()
 
 setup(
@@ -14,17 +14,17 @@
     # Versions should comply with PEP440.  For a discussion on single-sourcing
     # the version across setup.py and the project code, see
     # http://packaging.python.org/en/latest/tutorial.html#version
-    version='1.2.0',
+    version='0.0.1dev',
 
-    description='A sample Python project',
+    description='Kalman filtering and optimal estimation library',
     long_description=long_description,
 
     # The project's main homepage.
-    url='https://github.com/pypa/sampleproject',
+    url='https://github.com/rlabbe/filterpy',
 
     # Author details
-    author='The Python Packaging Authority',
-    author_email='pypa-dev@googlegroups.com',
+    author='Roger Labbe',
+    author_email='rlabbejr@gmail.com',
 
     # Choose your license
     license='MIT',
@@ -38,7 +38,7 @@
         'Development Status :: 3 - Alpha',
 
         # Indicate who your project is intended for
-        'Intended Audience :: Developers',
+        'Intended Audience :: Developers, Engineers',
         'Topic :: Software Development :: Build Tools',
 
         # Pick your license as you wish (should match "license" above)
@@ -56,7 +56,7 @@
     ],
 
     # What does your project relate to?
-    keywords='sample setuptools development',
+    keywords='Kalman filters filtering  optimal estimation tracking',
 
     # You can just specify the packages manually here if your project is
     # simple. Or you can use find_packages().