v0.2.0

LearnAPI v0.2.0

Diff since v0.1.0

This is a major, breaking rewrite of the original tentative "proposal" represented by 0.1.0, which was never intended for widespread adoption. Noteworthy change affecting StatisticalMeasures.jl and StatisticalMeasuresBase.jl:

• (breaking) LiteralTarget has been renamed Point.

Merged pull requests:

• Major refactor based on Julia Discourse feedback (#25) (@ablaom)
• Fix documentation generation (#26) (@ablaom)
• fix issue with predict_output_type (#28) (@ablaom)
• Major revision to simplify; in particular, to remove obsfit, obspredict, obstransform (#30) (@ablaom)
• Add a perceptron classifier to examples (#31) (@ablaom)
• Add an observation-updatable density estimator to tests (#32) (@ablaom)
• improve coverage (#33) (@ablaom)
• Coverage upgrade (#34) (@ablaom)
• Remove data slurping fallbacks (#35) (@ablaom)
• doc and test tweaks (#36) (@ablaom)
• add test (#37) (@ablaom)
• Use "learner" instead of "algorithm" in docs and in variable names (#38) (@ablaom)
• Remove testing of patterns (#40) (@ablaom)
• Fix links to sample patterns (#41) (@ablaom)
• Add a global default verbosity (#42) (@ablaom)
• Fix doc link (#43) (@ablaom)
• Update docs. No new release (#44) (@ablaom)
• For a 0.2.0 release (#45) (@ablaom)

Closed issues:

• Don't use "model" to describe hyperparameter struct. (#12)
• Accounting for data objects that only iterate (#15)
• Add fit_transform? (#18)
• Add a user-changeable global variable for default verbosity (#20)
• Add link to discussion in julia discourse regarding what to call learning algorithm configuration structs. (#21)
• Add new KindOfProxy: LabelAmbiguousFuzzy (#29)
• adopt getobs/numobs? (#39)