logisPred: Prediction function for a fitted logistic classifier

In MarcoDVisser/rML: Machine Learning routines for R

Description

Classifies from variables X

Classifies from variables X

Usage

logisPred(theta, X, thres = 0.5, type = "response")

decPlot(theta, dX, X, y, thres = 0.5)

Arguments

theta	numeric vector of parameters
X	design matrix: an array of numeric variables or features including intercepts
thres	decision boundry
type	type of predictions on logit scale ("logit") or "reponse" or as classification ("class")? Defaults to response
theta	numeric vector of parameters
dX	design matrix: an array of numeric variables or features
X	original variables or features
y	original classification including intercepts
thres	decision boundry

Author(s)

Marco D. Visser

Marco D. Visser

Examples

X<-array(runif(2000),dim=c(1000,2))
y<-sqrt(2*X[,2]^2+2.4*X[,1]^2)<.5
designX<-mapFeat(X)
designX<-cbind(rep(1,100),designX)
theta<-rep(0,ncol(designX))
lambda<-1
logisCost(theta,designX,y,lambda)
logisGrad(theta,designX,y,lambda)
par<-optim(theta,logisCost,logisGrad,X=designX,y=y,lambda=.01,method="BFGS")$par
decPlot(par,designX,X,y,thres=.5)

MarcoDVisser/rML documentation built on May 7, 2019, 2:49 p.m.