plotClassProbs

From caret v4.34 by Max Kuhn

Plot Predicted Probabilities in Classification Models

This function takes an object (preferably from the function extractProb) and creates a lattice plot. If the call to extractProb included test data, these data are shown, but if unknowns were also included, these are not plotted

Keywords: hplot

Usage

plotClassProbs(object,
               plotType = "histogram",
               useObjects = FALSE,
               ...)

Arguments

object: an object (preferably from the function extractProb. There should be columns for each level of the class factor and columns named obs, pred, model (e.g. "rpart"
plotType: either "histogram" or "densityplot"
useObjects: a logical; should the object name (if any) be used as a conditioning variable?
...: parameters to pass to histogram or densityplot

Value

A lattice object. Note that the plot has to be printed to be displayed (especially in a loop).

Examples

data(mdrr)
set.seed(90)
inTrain <- createDataPartition(mdrrClass, p = .5)[[1]]

trainData <- mdrrDescr[inTrain,1:20]
testData <- mdrrDescr[-inTrain,1:20]

trainY <- mdrrClass[inTrain]
testY <- mdrrClass[-inTrain]

ctrl <- trainControl(method = "cv")

nbFit1 <- train(trainData, trainY, "nb",
                trControl = ctrl,
                tuneGrid = data.frame(.usekernel = TRUE))

nbFit2 <- train(trainData, trainY, "nb",
                trControl = ctrl,
                tuneGrid = data.frame(.usekernel = FALSE))


models <- list(para = nbFit2,
               nonpara = nbFit1)

predProbs <- extractProb(models,
                         testX = testData, 
                         testY = testY)

plotClassProbs(predProbs,
               useObjects = TRUE)
plotClassProbs(predProbs,
               subset = object == "para" & dataType == "Test")
plotClassProbs(predProbs,
               useObjects = TRUE,
               plotType = "densityplot",
               auto.key = list(columns = 2))

Documentation reproduced from package caret, version 4.34, License: GPL-2

Community examples

Looks like there are no examples yet.