1.3.1

• linear regression, gradient boosting regression and regression forests are production ready
• hypothesis testing improvements: anderson darling, chi square independence, conditional independence and goodness of fit
• run time improvements for quantiles

1.3.0

Merge pull request #139 from hsnks100/LBFGS_singleton

apply LBFGS's classes to singleton pattern because those have static …

1.2.1

This release is dedicated to various small improvements.
The improved functionality concentrates on hypothesis testing and distributions.
For this release we have normal, t and chi square distribution done.
Hypothesis testing done for z tests, some of t tests.
There is also an implementation for FFT (Fast Fourier Transform) which will be improved further with various features.

1.2.0

Production ready old features:

• pearson and spearman correlation
• L1Regression, L2Regression, ConstantRegression, RFit
• many var and frame filters: VFToIndex, VFToNumeric, map vars, onehotencoding
• roc, roc curves
• Row sampling have a better shape
• SMO algorithm for SVM
• Var range is totally improved, by allowing lambdas.
• Improve sz option with factor and offset enhancement.

Production ready new features:

• random projection filters
• CTree: minGain parameter enhancement, remove nominal terms feature, improve var selection to select variables without exhaustion enhancement, make numeric_binary default test instead numerik-skip, improve gini gain critorion
• started a gitbook manual for rapaio library

Experimental features were added:

• CorrGram

Other things:

• Input filters now are duplicated as part of the new instance called in ML algorithms.
• Project was mavenized.
• Numerous bug fixes.

v1.1

This release contains various improvements on all the places.
Main achievements:

• classifier is production ready with all implementations
• regression was re-factored at API level, but implementations are experimental
• graphical API is production ready with various implementations
  • tons of small improvements

v1.0.2

This release has final API completed for core and data packages.
All other classes marked as @deprecated are considered in experimental stage and could be subject to change in future.

v1.0.1

This release has final API completed for core and data packages.
All other classes marked as @deprecated are considered in experimental stage and could be subject to change in future.

v1.0.0-beta

de-deprecate distribition implementations