llama (version 0.6)

featureFilter: Feature filtering

Description

Take data produced by input and determine the most predictive features (with respect to the best algorithm).

Usage

featureFilter(filter, data)

Arguments

filter

the filtering algorithm to use. Must accept a formula of the values to predict and a data frame with features. Return value should be the list of selected features.

data

the data to use. The structure returned by input.

Value

• features
  list of the most predictive features.
• original_features
  the list of the orignal features.
• ...
  the original members of data. See input.

Details

Runs the specified feature filtering algorithm on the data set with all features. The subset of the features chosen by the filter are saved in the features member, the old features (i.e. the full feature set) in original_features. The data structure modified like this is returned.

Examples

library(FSelector)

data(satsolvers)
d = featureFilter(filter=cfs, satsolvers)
d$features