featureMixtureProportion: featureMixtureProportion(unknown,known)

In jstarling1/varstar: What the package does (short line)

Description

This function calls the emCalcMix function repeatedly, for each feature (column) in the data set given. Creates estimated class proportion vector for each feature. Data set must have 'class' as the first column.

Usage

featureMixtureProportion(unknown, known, p = rep(0.2, 5), maxiter = 1000)

Arguments

unknown	A data frame containing rows of observations, with unknown classes. Columns contain features. Defaults to NULL.
known	A data frame containing rows of observations, with known classes. The first column must be class. Other columns contain features. Defaults to NULL.
p	A vector containing initial class proportion estimates. Defaults to (.2,.2,.2,.2,.2).

Value

A list containing two objects: $phat and $convergence

$phat is a list containing one element for each feature in the data set.

A list containing one element for each feature in the data set.

Each $phat list element is a vector of k mixing proportions.

$convergence is a data frame containing details on the EM algorithm convergence.

Author(s)

Jennifer Starling

Examples

est_mixing_proportions <- featureMixtureProportion(unknown=df1,known=df2)

jstarling1/varstar documentation built on May 20, 2019, 2:12 a.m.