Home

Welcome to the neural-network-modeling wiki! title: "neural networks" author: "Rizwan Ali" date: "July 12, 2018”

#packages and libraries
install.packages("neuralnet") install.packages("e1071") library("e1071") library(neuralnet)

checking data misiing values and then scaling and spliting into train and testing set.

data<-read.csv(choose.files()) apply(data,2,function(x)sum(is.na(x))) good<-complete.cases(data) data<-data[good,]

apply(data,2,function(x)sum(is.na(x)))

index <- sample(1:nrow(data),round(0.75*nrow(data)))

train <- data[index,] test <- data[-index,]

maxs <- apply(data, 2, max) mins <- apply(data, 2, min) scaled <- as.data.frame(scale(data, center = mins, scale = maxs - mins))

train_ <- scaled[index,] test_ <- scaled[-index,]

#making formula

n <- names(train_) f <- as.formula(paste("y ~", paste(n[!n %in% "y"], collapse = " + "))) #applying neural network model and plotting it.

nn <- neuralnet(f,data=train_,hidden=c(4),linear.output=T) plot(nn)

#computing and descaling the values.

pr.nn <- compute(nn,test_[,2:6]) pr.nn_ <- pr.nn$net.result*(max(data$y)-min(data$y))+min(data$y) test.r <- (test_$y)*(max(data$y)-min(data$y))+min(data$y) #checking the mean square error.

MSE.nn <- sum((test.r - pr.nn_)^2)/nrow(test_) print(paste(MSE.nn)) "14.5235812267609"

#fitting the regression line.

plot(test.r,pr.nn_,col='blue',main='real vs predicted',pch=1,cex=1,type="p",xlab = "actual",ylab = "predicted") abline(0,1,col="black")

coefficient of determination.

actual_pred<-data.frame(pred<-pr.nn$net.result,actual<-test_$y) cor(actual_pred)

                     pred....pr.nn.net.result actual....test_.y

pred....pr.nn.net.result 1.0000000000 0.9990796284 actual....test_.y 0.9990796284 1.0000000000

rsq<-function(x) 1-(x/nrow(train_)-1)*var(train_$y) rsq(.0338) 1.10924576