#------------------
# Data Preparation
#------------------

#Read datasets
#Download the data from http://www.saedsayad.com/datasets/BikeRental.zip
train <- read.csv("bike_rental_train.csv")
test <- read.csv("bike_rental_test.csv")

#Rows and Cols
dim(train)
dim(test)

#Columns name
colnames(train)
colnames(test)

#Show  
head(train)
head(test)


#Rows and Cols
dim(train)
dim(test)

#Columns name
colnames(train)
colnames(test)

#Show  
head(train)
head(test)

#Scatter plot
pairs(~temp+humidity+windspeed+bike_rent_count, data=train, main="Scatterplot - train", col="darkgreen")
pairs(~temp+humidity+windspeed+bike_rent_count, data=test, main="Scatterplot - test", col="brown")



#--------------------------------------
# Support Vector Machines - Regression
#--------------------------------------
library(e1071)

#Train
model.SVR <- svm(bike_rent_count~., train)
summary(model.SVR)

#Residual plot
res.SVR = train$bike_rent_count-predict(model.SVR, newdata=train)
plot(train$temp, res.SVR, ylab="Residuals", xlab="Temperature", main="Residual Plot") 
abline(0, 0)

#Q-Q plot
stdres.SVR = scale(res.SVR)
qqnorm(stdres.SVR, ylab="Standardized Residuals", xlab="Normal Scores", main="QQ Plot") 
qqline(stdres.SVR)

#Test
pred.SVR <- predict(model.SVR, newdata=test)
err.SVR <- test$bike_rent_count - pred.SVR 
rmse.SVR <- sqrt(mean((err.SVR^2)))

#Errors histogram
hist(err.SVR, main="bike_rent_count", sub="(Actual-Predicted)", xlab="Error", breaks=10, col="darkred")