kotlin-statistics changelog

Changelog History

• v1.2.1 Changes

  December 27, 2018

  0️⃣ Made some fixes to the binByComparable() implementation so ends are not exclusive by default.

• v1.2.0 Changes

  December 27, 2018

  🚀 This release has a major correction to the binning API, but it required Kotlin-Statistics to implement its own Range interface until Kotlin stdlib will support open ranges.

  When binning continuous numerical types (e.g. Double, BigDecimal, or Float), you no longer provide a gapSize argument as the ranges in the bin model will be end-exclusive.

  

• v1.1.3 Changes

  July 11, 2018

  🚀 This release contains some practical features for randomness and weighted probabilities.

  Kotlin-Statistics now has a few helpful extensions to randomly sample elements from an Iterable<T> or Sequence<T>.

  • randomFirst() - Selects one random element but throws an error if no elements are found.
  • randomFirstOrNull() - Selects one random element but returns null if no elements are found.
  • random(n: Int) - Selects n random elements.
  • randomDistinct(n: Int) - Select n distinct random elements.

  Also available are WeightedCoin and WeightedDice utilities which can be helpful to assign probabilities to discrete true/false or T outcomes. This can be helpful for games, stochastic programming, simulation, and optimization.

  enum class Move { ATTACK, DEFEND, HEAL, RETREAT}fun main(args: Array\<String\>) { val gameDice = WeightedDice( Move.ATTACK to .60, Move.DEFEND to .20, Move.HEAL to .10, Move.RETREAT to .10 ) val nextMove = gameDice.roll() println(nextMove) }
  

• v1.1.2 Changes

  April 15, 2018

  🚀 This is a re-release to fix some deployment issues with Kotlin's stdlib dependencies.

• v1.1.1 Changes

  April 14, 2018

  🚀 This release includes some minor fixes to the NaiveBayesClassifier as well as some new operators that will return ranges.

  • Sequence<T>.toNaiveBayesClassifier()
  • Sequence<C: Comparable<C>>.range()
  • Iterable<C: Comparable<C>>.range()
  • Sequence<C: Comparable<C>>.rangeBy()
  • Iterable<C: Comparable<C>>.rangeBy()
  • Sequence<Int>.intRange()
  • Iterable<Int>.intRange()
  • Sequence<T>.intRangeBy()
  • Iterable<T>.intRangeBy()
  • Sequence<Long>.longRange()
  • Iterable<Long>.longRange()
  • Sequence<T>.longRangeBy()
  • Iterable<T>.longRangeBy()
  • Sequence<Double>.doubleRange()
  • Iterable<Double>.doubleRange()
  • Sequence<T>.doubleRangeBy()
  • Iterable<T>.doubleRangeBy()
  • Sequence<Double>.doubleRange()
  • Iterable<Double>.doubleRange()
  • Sequence<T>.doubleRangeBy()
  • Iterable<T>.doubleRangeBy()
  • Sequence<Float>.floatRange()
  • Iterable<Float>.floatRange()
  • Sequence<T>.floatRangeBy()
  • Iterable<T>.floatRangeBy()

• v1.1.0 Changes

  April 07, 2018

  🚀 This release has a new NaiveBayesClassifier, as well as some generic projection fixes for the Bin and BinModel types.

  👀 See README for examples.

  val nbc = bankTransactions.toNaiveBayesClassifier( featuresSelector = { it.memo.splitWords().toSet() }, categorySelector = { it.category } ) //TESTval input = BankTransaction(date = LocalDate.of(2018,3,31), amount = 13.99, memo = "NETFLIX VIDEO ON DEMAND #21" ) val result = nbc.predict(input.memo.splitWords().toSet()) println(result) // prints "ENTERTAINMENT"
  

• v1.0 Changes

  October 29, 2017

  🚀 I'm excited to announce that I've settled on a base set of features and released Kotlin-Statistics 1.0. There have been some major refactorings from the pre-1.0 versions.

  • Maven artifact is now called kotlin-statistics, not kotlinstatistics
  • Most operators have been consolidated to target Number, rather than overloads for every numeric input type.
  • 👀 Many higher order function parameters have been renamed. Please see README for examples.
  • ✅ Unit tests have been written for approximately 96% of the codebase.

  I've decided to focus this library primarily on statistics, and steer away from linear algebra and other data science tasks. There are a few exceptions I've decided to keep, such as clustering and linear regression. I'll conservatively expand on these non-statistical features possibly, but I may spin off other projects.

  👀 If you have any ideas, please contribute and I'll be happy to consider adding features. Thank you for the enthusiastic support and usage of this library. What started as an experiment to see if Kotlin could be idiomatic for analytical tasks ended up becoming much more than I anticipated. I hope these patterns continue to gain traction and help make Kotlin a platform for data science.

• v0.3.0 Changes

  July 08, 2017

  🚀 There are some exciting new features in this release, including clustering algorithms.

  👀 KMeans and other clustering algorithms are now included. See README for details.

  binMapper parameters are now called valueMapper.

  Apache Commons Math 3 is now a dependency.

  🚚 Operators related to Short types have been removed, deemed uncommon and out-of-scope.

  👍 On the roadmap, operators will get better Pair support, and I will explore idiomatic linear algebra with vertices and matrices.

• v0.2.1 Changes

  May 28, 2017

  🔨 This includes some breaking refactors to ensure naming consistency:

  • All references to bucket are now called bin, including parameter names
  • 👌 Support for Iterable and Sequence now exist for binByXXX() operators

• v0.1.1 Changes

  May 26, 2017

  🚀 This release includes fixes for binIntBy() and binLongBy() operators, which were adding one extra increment to the first bucket.