API Reference / cli /

select_features.py

Select Features Command - Rank and keep the most informative features.

Functions

Command

tuiml select-features

Line 15
tuiml select-features [OPTIONS]

Select the most informative features in a dataset.

Scores every feature against the target with the chosen method and reports the surviving feature names and their column indices. Filter methods (SelectKBestSelector, SelectPercentileSelector, SelectFprSelector, SelectThresholdSelector, VarianceThresholdSelector), correlation-based selection (CFSSelector) and wrapper search (WrapperSelector) are all supported. Use --k to fix how many features survive, --threshold for the score-cutoff methods, and --method-params for anything else.

Rank features by their correlation with the target:

bash 
$ tuiml select-features --data iris --target class --method CFSSelector

Keep the two highest-scoring features:

bash 
$ tuiml select-features --data d.csv --target y --method SelectKBestSelector --k 2

Search for a good subset with a wrapper method:

bash 
$ tuiml select-features --data d.csv --target y --method WrapperSelector

Options

--data
str required
Data file path or built-in dataset name
--target
str required
Target column name
--method
str required
Feature selection class name (e.g., SelectKBestSelector, CFSSelector)
--k
int
Number of top-scoring features to keep (SelectKBestSelector)
--threshold
float
Score cutoff for VarianceThresholdSelector or SelectThresholdSelector
--method-params
str
Additional method-specific parameters as a JSON object, e.g. '{"percentile": 25}'
--json-output
flag
Output raw JSON