# Import Dataset

Startup = read.csv(file.choose())

#Encoding Categorical variables

# "California" =1 "Florida"=2 "New York"=3

Startup$State = factor(Startup$State,

levels = levels(Startup$State),

labels = c(1,2,3))

#Splitting the dataset into training and test sets

library(caTools)

set.seed(123)

split = sample.split(Startup$Profit, SplitRatio = 0.8)

training_set = subset(Startup, split==T)

test_set = subset(Startup, split==F)

#Fitting multiple linear regression to training set

regressor = lm(formula = Profit~., data=training_set)

summary(regressor)

#predicting test results

Y_pred = predict(regressor,newdata = test_set)

#Building optimal model using Backword Elemination

summary(regressor)

regressor = lm(formula = Profit~R.D.Spend+Administration+Marketing.Spend+State,

data=Startup)

summary(regressor)

#State3 has highest P value and higher than SL of 0.05, so remove it

regressor = lm(formula = Profit~R.D.Spend+Administration+Marketing.Spend,

data=Startup)

summary(regressor)

#Administration has highest P value and higher than SL of 0.05, so remove it

regressor = lm(formula = Profit~R.D.Spend+Marketing.Spend,

data=Startup)

summary(regressor)

#Marketing Spend has highest P value and higher than SL of 0.05, so remove it

regressor = lm(formula = Profit~R.D.Spend,

data=Startup)

summary(regressor)

regressor = lm(formula = Profit~R.D.Spend+Marketing.Spend,

data=training_set)

Y_pred = predict(regressor,test_set)

Y_pred