MLJBase.jl

MLJBase

Repository for developers that provides core functionality for theMLJ machinelearning framework.

Branch	Julia	Build	Coverage
master	v1
dev	v1

MLJ is a Juliaframework for combining and tuning machine learning models. Thisrepository provides core functionality for MLJ, including:

• completing the functionality for methods defined "minimally" inMLJ's light-weight model interfaceMLJModelInterface

• definition of machines and their associated methods, such asfit! and predict/transform. Serialization of machines,however, now lives inMLJSerialization.

• MLJ's model composition interface, including learningnetworks and pipelines

• basic utilities for manipulating data

• an extension toDistributions.jlcalled UnivariateFinite for randomly sampling labeledcategorical data

• a small interface for resampling strategies and implementations, including CV(), StratifiedCV and Holdout

• methods for performance evaluation, based on those resampling strategies

• one-dimensional hyperparameter range types, constructors andassociated methods, for use withMLJTuning

• a smallinterfacefor performance measures (losses and scores), enabling theintegration of theLossFunctions.jllibrary, user-defined measures, as well as about forty nativelydefined measures.

Previously MLJBase provided the model interface for integrating thirdparty machine learning models into MLJ. That role has now shifted tothe light-weightMLJModelInterfacepackage.