sparklightautoml.pipelines.features

Pipelines for features generation.

Base Classes

SparkFeaturesPipeline	Abstract class.
SparkTabularDataFeatures	Helper class contains basic features transformations for tabular data.
SparkEmptyFeaturePipeline	This class creates pipeline with SparkNoOpTransformer
SparkNoOpTransformer	This transformer does nothing, it just returns the input dataframe unchanged.
FittedPipe

Feature Pipelines for Boosting Models

SparkLGBSimpleFeatures	Creates simple pipeline for tree based models.
SparkLGBAdvancedPipeline	Create advanced pipeline for trees based models.

Feature Pipelines for Linear Models

SparkLinearFeatures

Creates pipeline for linear models and nnets.

Utility Functions

build_graph

Fill dict that represents graph of estimators and transformers