library(caret)

filename_train     <- "D:\\dat_train_final_experimental1.csv"
rdata_fronttest    <- "D:\\dat_test_features_experimental.RData"


trainData <- read.csv(filename_train, header=TRUE, sep=",", na.strings=c(".", "NA", "", "?"), strip.white=TRUE, encoding="UTF-8", check.names=FALSE)
trainData[, ncol(trainData)] <- factor(trainData[, ncol(trainData)])


if(!exists("predictorMask")){
	predictorMask <- c(FALSE,FALSE,FALSE,TRUE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,TRUE,FALSE,TRUE,FALSE,FALSE,FALSE,FALSE,FALSE,TRUE,TRUE,FALSE,FALSE,TRUE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,TRUE,FALSE,FALSE,FALSE,TRUE,TRUE,TRUE,TRUE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,FALSE,TRUE,FALSE,TRUE,FALSE,FALSE,FALSE,TRUE,FALSE,TRUE,FALSE,FALSE,FALSE,FALSE,TRUE,FALSE,TRUE,TRUE,TRUE,FALSE,FALSE,TRUE,FALSE,TRUE,FALSE,FALSE,FALSE,FALSE,TRUE,FALSE,TRUE,TRUE,FALSE,FALSE,FALSE,FALSE,TRUE,TRUE,FALSE,TRUE,FALSE,FALSE,TRUE,TRUE,TRUE,TRUE,TRUE,FALSE,TRUE,FALSE,FALSE,TRUE,FALSE,FALSE,TRUE,FALSE,TRUE,FALSE,FALSE)
}

if(!exists("nnetObj")){
	nnetObj <- train(x = trainData[, predictorMask],
					 y = trainData[, ncol(trainData)],
					 method = "nnet",
					 trControl = trainControl(method = "repeatedcv", p = 0.5),
					 preProcess = c("range"),
					 verbose = TRUE,
					 # nnet parameters:
					 MaxNWts=10000000,
					 trace=FALSE,
					 maxit=1000000,
					 skip = FALSE
					 )
}

predictResultsForTrain <- predict(nnetObj, newdata=trainData[, predictorMask])
predictResultsForTrainNumeric <- as.numeric(levels(predictResultsForTrain))[predictResultsForTrain]

"Accuracy on train when prediction != 0"
mean(predictResultsForTrainNumeric[which(predictResultsForTrainNumeric != 0)] == trainData[, ncol(trainData)][which(predictResultsForTrainNumeric != 0)])

"Accuracy on train on all examples"
mean(predictResultsForTrain == trainData[, ncol(trainData)])


load(rdata_fronttest)

trainDataFT <- dat_test_features
predictorMaskFT <- c(predictorMask, FALSE)#New column "symbol" which does not exist in train data
predictResultsForFronttest <- predict(nnetObj, newdata=trainDataFT[,predictorMaskFT])
predictResultsForFronttestNumeric <- as.numeric(levels(predictResultsForFronttest))[predictResultsForFronttest]

fronttestTargetClasses <- trainDataFT[,ncol(trainDataFT)]
fronttestTargetClasses[which(fronttestTargetClasses>0)] <- 1
fronttestTargetClasses[which(fronttestTargetClasses<0)] <- -1

"Accuracy on fronttest when prediction != 0"
mean(predictResultsForFronttestNumeric[which(predictResultsForFronttestNumeric != 0)] == fronttestTargetClasses[which(predictResultsForFronttestNumeric != 0)])

"Accuracy on fronttest on all examples"
mean(predictResultsForFronttest == factor(fronttestTargetClasses))

"Confusion matrix for train data"
confusionMatrix(predictResultsForTrain, trainData[, ncol(trainData)])$table

"Confusion matrix for fronttest data"
confusionMatrix(predictResultsForFronttest, factor(fronttestTargetClasses))$table