predict.matchFeat: Match New Feature Vectors To Existing Clusters

Description

predict method for class "matchFeat"

Usage

# S3 method for matchFeat
predict(object, newdata, unit = NULL, ...)

Value

A list of class matchFeat with fields

sigma: best matching as set of permutations (\((m,n)\) matrix)
cluster: best matching as cluster indicators (\((m,n)\)-matrix)
objective: minimum objective value
mu: mean vector for each class/label (\((p,m)\) matrix)
V: covariance matrix for each class/label (\((p,p,m)\) array if equal.variance is FALSE, \((p,p)\) matrix otherwise
call: function call

Arguments

object: an object of class "matchFeat".
newdata: new dataset of feature vectors
unit: unit labels for new data. Only necessary if newdata is a matrix
...: for compatibility with the generic predict method; argument not currently used.

Details

The function predict.matchFeat finds the best matching for new feature vectors relative to an existing set of cluster/class centers. If codeobject results from a call to match.gaussmix, the same function is used for prediction (with fixed mean vectors and covariance matrices). In other cases, the function match.template is used for prediction.

Examples

Run this code

data(optdigits)
train.result <- match.bca(optdigits$x[1:900,], optdigits$unit[1:900])  
test.result <- predict(train.result, optdigits$x[901:1000,], optdigits$unit[901:1000])
test.result