Classing_cluster: Cluster variables by correlation
In Zelazny7/binnr: Scorecard modeling in R

Cluster variables by correlation

`keep`	logical indicating whether to include dropped variables in the correlation cluster analysis.
`bag.fraction`	sample the development data without replacement to cluster more quickly. The bag.fraction is multiplied by the number of observations.

the cluster function first performs weight-of-evidence substitution. Cluster returns a classing_cluster object which is a list containing two fields: correlation & cluster. The first is a correlation matrix for all of the variables in the classing. The second is an hclust object which is result of hierarchical clustering of the correlation matrix.

a classing_cluster object

Zelazny7/binnr documentation built on May 10, 2019, 1:55 a.m.