Phd Report6 Training models based on dmax database

in this part,we build and tune 4 models(svm,neurons network,randomForest,Linear least squares) seperately by using package “caret”,based on dmax database, with 5 numbers and 5 repeats of leave-one-out cross validation

load("/home/bing/training/dmax/dmaxTrainingAndTesting.RData")
setwd("/home/bing/training/dmax")
library(doMC)
library(kernlab)
library(caret)
registerDoMC(cores = 3)
# svm RBF kernel
dmax.cvcontrol <- trainControl(method = "LOOCV", number = 5, repeats = 5)
if (file.exists("dmax.svmFit.RData")) {
    load("dmax.svmFit.RData")
} else {
    dmax.svmFit <- train(dmaxInputsTrain, dmaxTargetTrain, method = "svmRadial", 
        tuneLength = 4, trControl = dmax.cvcontrol, scaled = TRUE)
    save(dmax.svmFit, file = "dmax.svmFit.RData")
}
# neural networks
if (file.exists("dmax.nnetFit.RData")) {
    load("dmax.nnetFit.RData")
} else {
    nnet.grid <- expand.grid(.size = c(7:15), .decay = c(1e-04, 2e-04, 0.005, 
        0.01))
    dmax.nnetFit <- train(dmaxInputsTrain, dmaxTargetTrain, method = "nnet", 
        trControl = dmax.cvcontrol, tuneGrid = nnet.grid)
    save(dmax.nnetFit, file = "dmax.nnetFit.RData")
}
# random Forests
if (file.exists("dmax.rfFit.RData")) {
    load("dmax.rfFit.RData")
} else {
    library(randomForest)
    dmax.rfFit <- train(dmaxInputsTrain, dmaxTargetTrain, method = "rf", trControl = dmax.cvcontrol, 
        tuneLength = 4)
    save(dmax.rfFit, file = "dmax.rfFit.RData")
}
# Linear least squares
if (file.exists("dmax.lmFit.RData")) {
    load("dmax.lmFit.RData")
} else {
    dmax.lmFit <- train(dmaxInputsTrain, dmaxTargetTrain, method = "lm", trControl = dmax.cvcontrol, 
        tuneLength = 4)
    save(dmax.lmFit, file = "dmax.lmFit.RData")
}

Eorrors And Plot

# the function to caculate the model errors
modelErrors <- function(predicted, actual) {
    sal <- vector(mode = "numeric", length = 3)
    names(sal) <- c("MAE", "RMSE", "RELE")
    meanPredicted <- mean(predicted)
    meanActual <- mean(actual)
    sumPred <- sum((predicted - meanPredicted)^2)
    sumActual <- sum((actual - meanActual)^2)
    n <- length(actual)
    p3 <- vector(mode = "numeric", length = n)
    for (i in c(1:n)) {
        if (actual[i] == 0) {
            p3[i] <- abs(predicted[i])
        } else {
            p3[i] <- ((abs(predicted[i] - actual[i]))/actual[i])
        }
    }
    sal[1] <- mean(abs(predicted - actual))
    sal[2] <- sqrt(sum((predicted - actual)^2)/n)
    sal[3] <- mean(p3)
    sal
}
# prediction of svm,nnet,linearLeatSquare and randomForest models and plot
models <- list(svm = dmax.svmFit, nnet = dmax.nnetFit, linearLeatSquare = dmax.lmFit, 
    randomForest = dmax.rfFit)
dmax.preValues <- extractPrediction(models, testX = dmaxInputsTest, testY = dmaxTargetTest)
plotObsVsPred(dmax.preValues)

# build a function to predict differnet models and calculate the errors of
# those models
dmax.error <- function(model) {
    pd <- predict(model, newdata = dmaxInputsTest)
    modelErrors(pd, dmaxTargetTest)
}
rf.error <- dmax.error(dmax.rfFit)
svm.error <- dmax.error(dmax.svmFit)
nnet.error <- dmax.error(dmax.nnetFit)
lm.error <- dmax.error(dmax.lmFit)
errorAll <- rbind(svm.error, nnet.error, lm.error, rf.error)
errorAll

##               MAE   RMSE   RELE
## svm.error  0.1076 0.1436 0.4531
## nnet.error 0.1073 0.1417 0.4614
## lm.error   0.1116 0.1453 0.4787
## rf.error   0.1046 0.1369 0.4547