Multiple Linear Regression Predict Tip

Tip_prediction_multiple_linear_regression.r

# Multiple Linear Regression

# Importing the dataset
dataset = read.csv('tip.csv')

# Encoding categorical data, in this case country 
dataset$Country = factor(dataset$Country,
                       levels = c('USA', 'England', 'France', 'Israel', 'Italy'),
                       labels = c(1, 2, 3, 4, 5))

# install.packages('caTools')
# Load `caTools`
library(caTools)
# Set some random seed 
set.seed(123)

split = sample.split(dataset$Tip, SplitRatio = 0.8)
training_set = subset(dataset, split == TRUE)
test_set = subset(dataset, split == FALSE)


# Fitting Multiple Linear Regression to the Training set
regressor = lm(formula = Tip ~ .,
               data = training_set)

# Predicting the Test set results
y_pred = predict(regressor, newdata = test_set)

# By this point we should run `summary(regressor)`
# In order to check the p-value 
#Coefficients:
#              Estimate Std. Error t value Pr(>|t|)    
#(Intercept)   28.32170   15.30406   1.851 0.077702 .  
#WaiterAge      0.60355    0.27501   2.195 0.039042 *  
#GroupSize     -2.90429    2.91919  -0.995 0.330603    
#DinersAvgAge  -0.07064    0.24696  -0.286 0.777532    
#CostOfMeal     0.09314    0.04918   1.894 0.071446 .  
#Country2     -48.60033   13.15088  -3.696 0.001263 ** 
#Country3     -48.62730   12.46118  -3.902 0.000765 ***
#Country4     -21.03267   13.19317  -1.594 0.125156    
#Country5     -38.42068   15.17851  -2.531 0.019020 *  
#---
# Let's get rid of `DinersAveAge

# Taking out DinersAvgAge
regressor2 = lm(formula = Tip ~ WaiterAge + GroupSize + CostOfMeal + Country,
                data = training_set)
y_pred2 = predict(regressor, newdata = test_set)

# Again let's run `summary(regressor2)`
# WaiterAge     0.60737    0.26915   2.257 0.033841 *  
# GroupSize    -2.72674    2.79491  -0.976 0.339408    
# CostOfMeal    0.09151    0.04786   1.912 0.068410 .  
# Country2    -50.10668   11.80735  -4.244 0.000307 ***
# Country3    -49.63997   11.70673  -4.240 0.000309 ***
# Country4    -22.21611   12.27500  -1.810 0.083404 .  
# Country5    -39.49177   14.41273  -2.740 0.011666 *  
# Let's get rid of Group Size
# Taking out GroupSize
regressor3 = lm(formula = Tip ~ WaiterAge + CostOfMeal + Country,
                data = training_set)
y_pred3 = predict(regressor, newdata = test_set)
# Running `summary(regressor3)`
# WaiterAge     0.57197    0.26643   2.147 0.042124 *  
# CostOfMeal    0.04684    0.01392   3.366 0.002564 ** 
# Country2    -48.31559   11.65204  -4.147 0.000364 ***
# Country3    -48.69739   11.65508  -4.178 0.000336 ***
# Country4    -21.92541   12.25906  -1.789 0.086327 .  
# Country5    -34.71975   13.54374  -2.564 0.017050 *  

# We're looking for p-balue <= 0.05 so let's take out `Country`
# Taking out Country
regressor4 = lm(formula = Tip ~ WaiterAge + CostOfMeal,
                data = training_set)
y_pred4 = predict(regressor, newdata = test_set)

# And we got
# WaiterAge    0.37441    0.33924   1.104   0.2791  
# CostOfMeal   0.03703    0.01795   2.063   0.0485 *

# We can learn that the biggest factor of a tip is the cost of the `meal` but the second factor is 
# the waiter age
# Now we can Run our model on our test dataset and compare it to our train dataset