ML Using R Decision Tree Regression

# Import data set
dataset <-  read.csv("G:\\RStudio\\udemy\\ml\\Machine Learning AZ\\Part 2 - Regression\\Section 8 - Decision Tree Regression\\Decision_Tree_Regression\\Position_Salaries.csv")
# We want to use only columns 2 and 3
dataset <-  dataset[2:3]
head(dataset)

# Polynomial Regression
# taking care of missing values
# Test for missing values
sum(is.na(dataset$Level))

[1] 0

sum(is.na(dataset$Salary))

[1] 0

There is no missing values in the dataset.

The while the next step should be to split the dataset, since this dataset is small, we will use the full dataset.

# feature scaling
# in this case for polynomial, not needed for feature scaling. 

# Fitting the  regression model
# add a new column in the dataframe
# install.packages("rpart")
library(rpart)

package <U+393C><U+3E31>rpart<U+393C><U+3E32> was built under R version 3.3.3

regressor <-rpart(formula = Salary ~., data = dataset)
summary(regressor)

Call:
rpart(formula = Salary ~ ., data = dataset)
  n= 10 

    CP nsplit rel error xerror xstd
1 0.01      0         1      0    0

Node number 1: 10 observations
  mean=249500, MSE=8.066225e+10 

# predicting a new result 
y_pred <-  predict(regressor, data.frame(Level = 6.5))
y_pred

[1] 249500

# Plot
library(ggplot2)
ggplot() +
  geom_point(aes(x = dataset$Level, y = dataset$Salary), colour = "red") +
  geom_line(aes(x = dataset$Level, y = predict(regressor, newdata = dataset)), colour = "blue")+
  ggtitle("Truth or Bluff (Regression Model)") +
  xlab("Levels") +
  ylab("Salary")

# visualizing polynomial regression
regressor <-rpart(formula = Salary ~., data = dataset, control = rpart.control(minsplit = 1))
summary(regressor)

Call:
rpart(formula = Salary ~ ., data = dataset, control = rpart.control(minsplit = 1))
  n= 10 

          CP nsplit  rel error   xerror      xstd
1 0.77638626      0 1.00000000 1.234568 0.7835133
2 0.15496716      1 0.22361374 1.148378 0.7931845
3 0.05217357      2 0.06864658 1.120316 0.7968327
4 0.01000000      3 0.01647301 1.120316 0.7968327

Variable importance
Level 
  100 

Node number 1: 10 observations,    complexity param=0.7763863
  mean=249500, MSE=8.066225e+10 
  left son=2 (8 obs) right son=3 (2 obs)
  Primary splits:
      Level < 8.5 to the left,  improve=0.7763863, (0 missing)

Node number 2: 8 observations,    complexity param=0.05217357
  mean=124375, MSE=6.921484e+09 
  left son=4 (6 obs) right son=5 (2 obs)
  Primary splits:
      Level < 6.5 to the left,  improve=0.7600316, (0 missing)

Node number 3: 2 observations,    complexity param=0.1549672
  mean=750000, MSE=6.25e+10 
  left son=6 (1 obs) right son=7 (1 obs)
  Primary splits:
      Level < 9.5 to the left,  improve=1, (0 missing)

Node number 4: 6 observations
  mean=82500, MSE=1.38125e+09 

Node number 5: 2 observations
  mean=250000, MSE=2.5e+09 

Node number 6: 1 observations
  mean=500000, MSE=0 

Node number 7: 1 observations
  mean=1000000, MSE=0 

# predicting a new result 
y_pred <-  predict(regressor, data.frame(Level = 6.5))
y_pred

     1 
250000 

library(ggplot2)
ggplot() +
  geom_point(aes(x = dataset$Level, y = dataset$Salary), colour = "red") +
  geom_line(aes(x = dataset$Level, y = predict(regressor, newdata = dataset)), colour = "blue")+
  ggtitle("Truth or Bluff (Regression Model)") +
  xlab("Levels") +
  ylab("Salary")

# visualizing in higher resolution
x_grid = seq(min(dataset$Level), max(dataset$Level), 0.1 )
ggplot() +
  geom_point(aes(x = dataset$Level, y = dataset$Salary), colour = "red") +
  geom_line(aes(x = x_grid, y = predict(regressor, newdata = data.frame(Level = x_grid))), colour = "blue")+
  ggtitle("Truth or Bluff (Regression Model)") +
  xlab("Levels") +
  ylab("Salary")