emuR (version 0.1.7)

train: Train a Gaussian Model

Description

Trains a Gaussian Model

Usage

train(x, lab = rep("x", nrow(x)))

Arguments

x

A data vector or matrix.

lab

A vector of labels parallel to x. If missing, all data is assumed to be from the same class.

Value

• A structure with the following components:
• label
  The unique labels in lab.
• means
  The means for each dimension per unique label.
• cov
  The combined covariance matrixes for each unique label. The matrixes are joined with rbind. If the input data is one-dimensional, this is just the standard deviation of the data.
• invcov
  The combined inverse covariance matrixes for each unique label. The matrixes are joined with rbind. If the input data is one-dimensional, this is just the reciprocal of the standard deviation of the data.

Details

This function is used to train a gaussian model on a data set. The result can be passed to either the mahal or bayes.lab functions to classify either the training set (x) or a test set with the same number of dimensions. Train simply finds the mean and inverse covariance matrix/standard deviation for the data corresponding to each unique label in labs.

See Also

mahal, bayes.lab, mahalplot, bayes.plot