LRMultiClass: Function that implements multi-class logistic regression.

In gargjhanvi/SSLLabelGuide: Classify what data points should be labelled(for training) based on budget constraints and very small or no labelled data for semi supervised learning, predict the labels of unlabelled data and determine the confidence of labelling them

Description

Function that implements multi-class logistic regression.

Usage

LRMultiClass(X, y, Xt, yt)

Arguments

X	n x p training data, 1st column should be 1s to account for intercept
y	a vector of size n of class labels, from 0 to K-1
Xt	ntest x p testing data, 1st column should be 1s to account for intercept
yt	a vector of size ntest of test class labels, from 0 to K-1

Value

A list of two elements,

beta	p x K matrix of estimated beta values after numIter iterations
error_train	(numIter + 1) length vector of training error % at each iteration (+ starting value)

Examples

X1 <- matrix(rnorm(50, -3, 1), 50, 1)
Y1 <- matrix(c(0), 50, 1)
X2 <- matrix(rnorm(50, 3, 1), 50, 1)
Y2 <- matrix(c(1), 50, 1)
X <- c(X1, X2)
Y <- c(Y1, Y2)
X <- as.matrix(X)
Y <- as.matrix(Y)
random <- sample(nrow(X))
X <- as.matrix(X[random, ]) # training data from two normal distributions
Y <- as.matrix(Y[random, ]) # class labels for training data
# creating test data
X1t <- matrix(rnorm(10, -3, 1), 10, 1)
Y1t <- matrix(c(0), 10, 1)
X2t <- matrix(rnorm(10, 3, 1), 10, 1)
Y2t <- matrix(c(1), 10, 1)
Xt <- c(X1t, X2t)
Yt <- c(Y1t, Y2t)
Xt <- as.matrix(Xt)
Yt <- as.matrix(Yt)
random <- sample(nrow(Xt))
Xt <- as.matrix(Xt[random, ]) # testing data from the same two normal distributions
Yt <- as.matrix(Yt[random, ]) # class labels for testing data
# adding columns of 1's to training and testing data
colX1 <- rep(1, nrow(X))
X <- cbind(colX1, X)
colXt1 <- rep(1, nrow(Xt))
Xt <- cbind(colXt1, Xt)
output <- LRMultiClass(X, Y, Xt, Yt)

gargjhanvi/SSLLabelGuide documentation built on Dec. 20, 2021, 9:48 a.m.