v1.0.1

Add support for Crystal 1.0.0.

v1.0.0

• Introduce DiscreteDistribution and ContinuousDistribution
• Add pdf and pmf to continuous and discrete distributions, respectively
• Add Bernoulli distribution
• Update Statistics.bin_count to be more flexible - see docs

v0.3.0

• Upgrade to Crystal 0.34.0

• Add support for data binning with Statistics.bin_count.

  sample = [0.5, 0.8, 0.85, 1.2, 2, 2.2, 2.8, 3.3, 4]
  
  Statistics.bin_count(sample, bins: 4, min: 0)
  # [{0.0, 3}, {1.0, 1}, {2.0, 3}, {3.0, 2}]

  You can check out the docs for more details.

v0.2.0

Parametrize Distribution in the return type of Distribution#rand.

This improves type resolution when mixing different distributions, for example in the expression

x = rand < 0.5 ? Normal.new(0.1, 0.02) : Uniform.new(0.2, 0.5)

x is resolved to the type Distribution(Float64).

v0.1.1

Make all the probability distributions inherit from a single abstract class Statistics::Distributions::Distribution.

This simplifies the signature of methods accepting different distributions.

Bonus: improve Distributions module documentation.

v0.1.0

The first release of statistics includes the following statistical functions

• mean, median, middle, mode
• std, var
• skew, kurtosis
• quantile
• frequency
• moment

It also includes the convenience function describe, to provide an overview of the main statistical measures for a given dataset.

Finally, in the Statistics::Distributions module, you will find classes to sample random numbers from the following statistical distributions

• Constant
• Exponential
• Normal
• Poisson
• Uniform