Feature Selection Help

Rank and select the most informative features (importance, SelectKBest, RFE, SFS and more). Produces a derived dataset with the selected features plus the target.

Requires: One dataset with a numeric target column.

Form fields accepted by this method's POST request

analysis_type=feature_selection, dataset_id, feature_selection_method, fs_estimator, fs_threshold, k, target_col

No dataset yet? Upload a file.
Supervised methods score each feature against this target.
Parameters & Options
k-best, importance, RFE and SFS select this many features; SelectPercentile treats it as a percentage (1-100).
Optional: fill missing numeric values before the analysis runs. Leave as default to keep the method's own handling.
Back to Methods