Feature reduction

Feature reduction operates on the preprocessed features. It allows you to reduce the dimension of the feature space in order to regularize your model or make it more interpretable.

This creates a LASSO model to predict the {{ isCausalPrediction() | targetRoleName}}, using 3-fold cross-validation to select the best value of the regularization term. Only the features with nonzero coefficients will be selected.

Regularization

The feature space dimension is reduced using Principal Component Analysis, only the top principal components are selected.
This method generates non-interpretable feature names as its output. The model may be performant, but some results may not be interpretable. Feature importance computation is also slower.

Maximum components

Proportion of variance to keep

Only the features most correlated ({{ isClassification() ? 'ANOVA F-score' : 'Pearson'}}) with the {{ isCausalPrediction() | targetRoleName}} will be selected. A threshold for minimum and maximum absolute correlation can be set.

Minimum absolute correlation

Maximum absolute correlation

Number of features to keep