sid_halve: Calculate goodness for 2-group attribute sets.

In bwrc/astrid-r: Automatic STructure IDentification

Description

fter ranking, the attributes are ordered so that those belonging to the same group are consecutive. Now calculate the accuracy for each grouping formed by splitting this sorted list of attributes into a tree with 2 groups, for each possible split (number of attributes - 1).

Usage

sid_halve(data_train, data_test, attr_rank, R = 10, classifier = svm,
  parallel = parallel)

Arguments

data_train	Training data.
data_test	Testing data.
attr_rank	Sorted list of attributes (from sid_sort).
R	Number of replications. Default is 10.
classifier	A classifier function, default is svm (from the e1071 package).
parallel	Calculate in parallel (Boolean, default is TRUE).

Value

The attributes, sorted.

bwrc/astrid-r documentation built on May 13, 2019, 9:08 a.m.