mlr (version 2.1)

getFilterValues: Calculates feature filter values.

Description

Calculates numerical filter values for all features. Look at package FSelector for details on the filter algorithms.

Currently only supports classification (C) and regression (R). Allowed feature types are abbreviated in table as numerics (N) and factors (F).

Available methods are: llll{ Method Tasks Feats Description linear.correlation R N Pearson's correlation between feature and target rank.correlation R N Spearman's correlation between feature and target information.gain C,R N,F Entropy-based information gain between feature and target gain.ratio C,R N,F Entropy-based gain ratio between feature and target symmetrical.uncertainty C,R N,F Entropy-based symmetrical uncertainty between feature and target chi.squared C,R N,F Chi-squared statistic of independence between feature and target random.forest.importance C,R N,F See importance relief C,R N,F RELIEF algorithm oneR C,R N,F OneR assocation rule mRMR.classic R N MRMR algorithm, see mRMR.classic from mRMRe package }

Usage

getFilterValues(task, method = "random.forest.importance", ...)

Value

[FilterValues].

See Also

Other filter: FilterResult, getFilterResult; FilterValues; filterFeatures; getFilteredFeatures; listFilterMethods; makeFilterWrapper