How to perform multiple linear regression in R? Explain with an example

data<- read.csv(file.choose()) # choose the 50_startup.csv data set

View(data)

summary(data)

Now we will find the correlation of the data

# 7. Find the correlation b/n Output and input

pairs(data)

data=data[,-4]

# 8. Correlation Coefficient matrix - Strength & Direction of Correlation

cor(data)

### Partial Correlation matrix - Pure Correlation b/n the varibles

#install.packages("corpcor")

library(corpcor)

cor2pcor(cor(data))

Now we will fit the model and rebuild model by removing each columns to see the change in accuracy.

# The Linear Model of interest with all the columns

model.data <- lm(Profit~.,data=Cars)

# Model based on RD and Adm

model.data1<-lm(Profit~R.D.Spend+Administration,data=Cars)

summary(model.data1)

# Model based on RD and Marketing

model.data2<-lm(Profit~R.D.Spend+Marketing.Spend,data=Cars)

summary(model.data2)

#Marketing and Administration are highly insignificant so we are removing

finalmodel<-lm(Profit~R.D.Spend,data=data)

summary(finalmodel)

Now we will evaluate the model to prove the assumptions that the errors should be normally distributed.

# Evaluate model LINE assumptions

plot(finalmodel)

hist(residuals(finalmodel)) # close to normal distribution