Skip to content

Latest commit

 

History

History
274 lines (198 loc) · 9.45 KB

CHANGES.md

File metadata and controls

274 lines (198 loc) · 9.45 KB
Raw

krangl Release History

v0.15

New Features

// read
df = DataFrame.readExcel("data.xlsx", sheetName = "sales")
df = DataFrame.readExcel("data.xlsx", cellRange = CellRangeAddress.valueOf("A1:D10"))

// write
df.writeExcel("results.xslx")
  • #95 Improved column type casts
dataFrameOf("foo")(1, 2, 3).addColumn("stringified_foo") { it["foo"].toStrings() }.schema()
> DataFrame with 3 observations
> foo              [Int]  1, 2, 3
> stringified_foo  [Str]  1, 2, 3

dataFrameOf("foo")("1", "2", "3").addColumn("parsed_foo") { it["foo"].toInts() }.schema()

> DataFrame with 3 observations
> foo         [Str]  1, 2, 3
> parsed_foo  [Int]  1, 2, 3
  • #99 Added filtering by list (similar to R's %in% operator)
irisData.filter { it["Species"].inList("setosa", "versicolor")  }

Bug Fixes

  • #84 Builder now supports mixed numbers in column
  • #96 & #94 Fixed bugs in join
  • #100 Improved SQL bindings
  • #99 Fixed median

v0.14

  • Fixed missing by values overhanging RHS in outer join (fixes #94)
  • Added addRow (via PR92 by LeandroC89
  • Added column type text to sql interface (fixes #72)

v0.13

Released: 2020-06-02

  • Added column transformation to calculate cumulative sum cumSum
sales
    .sortedBy("quarter")
    .addColumn("cum_sales" to { it["sold_units"].cumSum()})
  • Added column transformation pctChange to calculate percentage change between the current and a prior element. similar to pct_change in pandas (contributed by @amorphous1 in PR85)
sales
    .groupBy("product")
    .addColumn("sales_pct_change" to { it["sold_units"].pctChange() })
  • Added lead and lag (contributed by @amorphous1 in PR85)
sales
    .groupBy("product")
    .sortedBy("quarter")
    .addColumn("prev_quarter_sales" to { it["sold_units"].lag() })

  • Significantly improved join performance (contributed by @amorphous1 in PR85)

  • New: Extended bindRows API to combine data rowwise (see PR #77 by @CrystalLord)

val person1 = mapOf("person" to "James", "year" to 1996)
val person2 = mapOf("person" to "Anne", "year" to 1998)

emptyDataFrame().bindRows(person1, person2).print()

v0.12

internal release

v0.11

  • New: Added built-it support for Long columns (PR #69 by @davidpedrosa)

v0.10

Major:

  • New: summarizeAt for simplified column aggregations
  • New: setNames to replace column headers of a data-frame
  • New: Deparse Iterables more conveniently using lambdas in deparseRecords

Minor:

  • Fixed: Can not read csv-tables without header
  • Added option to skip lines in csv reader.
  • Fixed schema() should no throw memory exception (#53: )
  • Fixed DataFrame.readTSV default format (#56)
  • Added where() for conditional column creation (relates to #54)
  • Added writeTSV
  • Fixed grouping by Any columns
  • Added: toDoubleMatrix() helper extension method

v0.9.1

Major Enhancements

  • DataFrame.fromJson will now flatten nested json data

Minor

  • Added sum() extension for columns summaries/transformation
  • Added dataFrameOf() that accepts Iterable of names
  • Added bindRows() alias that accepts data frames as varargs
  • Added bindCols() extension for list of DataCol
  • Fill missing cells with NA in bindRows and bindCols
  • Resolve duplicated column names in bindCols()
  • Added new builder to create data-frame from DataFrameRow iterator
  • Added addRowNumber to add the row number as column to a data-frame
  • Fixed: Incorrect types in gathered columns

v0.9

Released 2018-04-11

Major Enhancements

  • Allow index access for column model (fixes #46): irisData[1][2]
  • Improved DataFrame.count to respect existing groupings and to simply count rows if no grouping is defined
  • Added moveLeft and moveRight to rearrange column order
  • Added nest and unnest to wrap columns into sub-tables and back
  • Added expand and complete to expand column value-sets into data-frames
  • Added function literal support for count and groupBy (fixes #48): irisData.groupByExpr{ it["Sepal.Width"] > 3 }
  • Added receiver context for sortBy lambdas with sorting specific API (fixes #44)

Improved data-frame rendering

  • Improved print()ing of data-frames and schema()ta to have better alignment and more formatting options
  • Print row numbers by default when using print (fixes #49)

Minor Enhancements

  • Renamed select2/remove2 to selectIf and removeIF
  • Fixed #39: Can not add scalar object as column
  • Started submodule for documentation
  • Hide columns in print after exceeding maximum line length (fixes #50)
  • Fixed #45: sleepData.sortedBy{ "order" } should fail with informative exception

v0.8

Released 2018-03-21

Major Enhancements

  • Added property unfolding df.unfold<Person("user", properties=listOf("address"))
  • Added text matching helper: irisData.filter{ it["Species"].isMatching{ startsWith("se") }} (fixes #21)
  • Added sortedByDescending and desc and added more sorting tests
  • Added More elegant object bindings via reflection. Example val objPersons : Iterable<User> = users.rowsAs<User>() (fixes #22)
  • Added compressed csv write support, configurable or by filename guessing

Minor Enhancements

  • More robust row to object conversion
  • Made List<Boolean?>.not() public
  • Use regex instead of string as separate separator
  • Replaced fixed temporary column names with uuids
  • Fixed incorrect coercion of incomplete inplace data to df
  • Added concat operator for string column arithmetics
  • Fixed arithmetic comparison operators
  • Added beakerx display adapter

v0.7

Released 2018-03-14

Major Enhancements

  • Allow specifying column types when reading csv data (Thanks to LeanderG for providing the PR)
  • Added groupedBy to provide distinct set of grouping tuples as data-frame
  • Read support for URLs (Example DataFrame.readCSV("https://git.io/vxks7").glimpse())
  • Added basic read/write support for JSON data
  • Added generic collection conversion Iterable<Any>.asDataFrame() via reflection (fixes #24)

Incompatible API changes

  • Renamed structure to columnTypes
  • Renamed all table read function from .from* to .read*
  • Fixed #29: mapNonNull should use parameter and not receiver

Minor Enhancements

  • Namespace cleanup to hide internal helpers
  • Bundled irisData
  • Enhanced: DataCol.toDouble() should work for int columns as well (same vv)
  • Added MIT License
  • Use iterable instead of list for object conversions

v0.6

Released: 2017-11-11

  • More idiomatic API mimicking kotlin stdlib where possible
  • Added DataFrame.remove to drop columns from data-frames
  • Added DataFrame.addColumn to add column from data-frames
  • Added DataFrame.sortBy(TableFormula)
  • Added DataFrame.filterByRow
  • Reworked column selector API
  • Changed column expression API from Any to a constrained set of support types
  • Fixed issues when combining columns of different types (e.g. DoubleCol + IntCol
  • Dropped most unary operators

v0.5

Skipped.

v0.4

released on 2017-4-12

New Features

  • spread()-gather() support for elegant data reshaping (fixes #2)
  • Improve reshaping functionality by adding unite and separate (fixes #9)
  • Added sampleFrac() and sampleN() for random sub-sampling of data-frames (either with or without replacement)

Important Bug Fixes

  • mutate() can now change existing columns without altering column positions

Other

  • New property accessor DataFrame.cols to access all columns of a data-frame
  • Incremented kotlin version to 1.1

v0.3

Initial Release

  • Implement all dplyr core verbs
  • Implement all join types
  • Table write support using csv-commons wrapper
  • Extensive unit test coverage =
  • TravisCI integration
  • Support for count() and distinct()
  • Basic benchmarking framework (without jvm usage)