Data Mining - Predictive Modelling
Classification using Decision
Trees
Data Mining - Predictive Modelling
Classification using Decision Trees
Updated by Uwe H Kaufmann on 22 Jan 2025, 22:07
Updated by Uwe H Kaufmann on 22 Jan 2025, 22:07
- 1 Load Libraries
- 2 Classification and Regression Tree
Using CART
- 2.1 Get Data to Classify Whether or not a Car Insurance Claim was Fraudulent
- 2.2 Calculate Decision Tree
- 2.3 Get Data for Predicting Whether To Play Tennis
- 2.4 Use Predictors Whether To Play Tennis to Develop Decision Tree - Using Recursive Partitioning (RPart)
- 2.5 Get Data to Determine Whether Customer Should be Classified as Credit Risk
- 2.6 Use Predictors to Determine Whether Customer Should be Classified as Credit Risk
- 3 Classification and Regression Tree Using Conditional Inference Tree (CTREE, C4.5 Method)
- 4 Classification and Regression Tree Using CART
- 5 Classification and Regression Tree Using Conditional Inference Tree (CTREE, C4.5 Method)
- 6 Classification and Regression Tree
Using Conditional Inference Tree (C5.0, C5.0 Method)
- 6.1 Calculate Decision Tree Using C5.0
- 6.2 Plot Decision Tree Using C5.0
- 6.3 Get Data to Determine Whether Customer Should be Classified as Credit Risk
- 6.4 Use Predictors to Determine Whether Customer Should be Classified as Credit Risk Using C5.0
- 6.5 Plot Decision Tree to Determine Whether Customer Should be Classified as Credit Risk Using C5.0
- 7 Assignment
- 8 Plot Decision Tree Using Partykit
- 9 Classification Using Random Forest
1 Load Libraries
# Loading necessary packages
## Markdown Update
if(! "rmarkdown" %in% installed.packages()) { install.packages("rmarkdown", dependencies = TRUE) }
library(rmarkdown)
## Loading other packages if not available
if(! "readxl" %in% installed.packages()) { install.packages("readxl", dependencies = TRUE) }
library(readxl)
if(! "tidyr" %in% installed.packages()) { install.packages("tidyr", dependencies = TRUE) }
library(tidyr)
if(! "zoo" %in% installed.packages()) { install.packages("zoo", dependencies = TRUE) }
library(zoo)
if(! "car" %in% installed.packages()) { install.packages("car", dependencies = TRUE) }
library(car)
# Global Settings
options(digits = 4)
options(scipen = 999)
# Setting working Directory
setwd("~/AC UNI-ORG/AB SIM/GDBA/R")
# Make date string
today <- format(as.Date(Sys.time(), tz = "Asia/Singapore"), format = "%y%m%d")2 Classification and Regression Tree Using CART
2.1 Get Data to Classify Whether or not a Car Insurance Claim was Fraudulent
# Load Library if not available
if(! "rpart" %in% installed.packages()) { install.packages("rpart", dependencies = TRUE) }
library(rpart)
if(! "rpart.plot" %in% installed.packages()) { install.packages("rpart.plot", dependencies = TRUE) }
library(rpart.plot)
# Input Data Set
train <- data.frame(
ClaimID = c(1,2,3),
RearEnd = c(TRUE, FALSE, TRUE),
Fraud = c(TRUE, FALSE, TRUE)
)
# Show Data
train## ClaimID RearEnd Fraud
## 1 1 TRUE TRUE
## 2 2 FALSE FALSE
## 3 3 TRUE TRUE# Write data to working directory
write.csv(train, file = "ClaimTrain.csv")2.2 Calculate Decision Tree
# Calculate the Tree
mytree <- rpart(
Fraud ~ RearEnd,
data = train,
method = "class",
minsplit = 2,
minbucket = 1
)
print(mytree)## n= 3
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 3 1 TRUE (0.3333 0.6667)
## 2) RearEnd< 0.5 1 0 FALSE (1.0000 0.0000) *
## 3) RearEnd>=0.5 2 0 TRUE (0.0000 1.0000) *# Plot the Tree
cat("\n\nThe following information is shown in each node:\n")##
##
## The following information is shown in each node:cat("- Node Number\n")## - Node Numbercat("- Predicted class\n")## - Predicted classcat("- Predicted probability of this class\n")## - Predicted probability of this classcat("- Percentage of Observations in the node\n")## - Percentage of Observations in the noderpart.plot(mytree, main = "Classification Tree To Identify Insurance Fraud",
under = FALSE,
type = 2,
extra = 106,
clip.right.labs = TRUE, shadow.col = "gray", # shadows under the node boxes
nn = TRUE,
fallen.leaves = TRUE,
digits = 3,
box.palette = "RdGn"
)
2.3 Get Data for Predicting Whether To Play Tennis
# Download Data from URL2.4 Use Predictors Whether To Play Tennis to Develop Decision Tree - Using Recursive Partitioning (RPart)
# Change Play Tennis Column
PlayTennis$PlayTennis <- ifelse(PlayTennis$PlayTennis == "Yes", "Play", "Don't Play")
# Show Data
PlayTennis## Day Outlook Temperature Humidity Wind PlayTennis
## 1 1 Sunny Hot High Weak Don't Play
## 2 2 Sunny Hot High Strong Don't Play
## 3 3 Overcast Hot High Weak Play
## 4 4 Rainy Mild High Weak Play
## 5 5 Rainy Cool Normal Weak Play
## 6 6 Rainy Cool Normal Strong Don't Play
## 7 7 Overcast Cool Normal Strong Play
## 8 8 Sunny Mild High Weak Don't Play
## 9 9 Sunny Cool Normal Weak Play
## 10 10 Rainy Mild Normal Weak Play
## 11 11 Sunny Mild Normal Strong Play
## 12 12 Overcast Mild High Strong Play
## 13 13 Overcast Hot Normal Weak Play
## 14 14 Rainy Mild High Strong Don't Play# Calculate the Tree
TennisTree <- rpart(
PlayTennis ~ Outlook + Temperature + Wind + Humidity,
data = PlayTennis,
method = "class",
minsplit = 2,
minbucket = 1
)
print(TennisTree)## n= 14
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 14 5 Play (0.3571 0.6429)
## 2) Outlook=Rainy,Sunny 10 5 Don't Play (0.5000 0.5000)
## 4) Humidity=High 5 1 Don't Play (0.8000 0.2000)
## 8) Outlook=Sunny 3 0 Don't Play (1.0000 0.0000) *
## 9) Outlook=Rainy 2 1 Don't Play (0.5000 0.5000)
## 18) Wind=Strong 1 0 Don't Play (1.0000 0.0000) *
## 19) Wind=Weak 1 0 Play (0.0000 1.0000) *
## 5) Humidity=Normal 5 1 Play (0.2000 0.8000)
## 10) Wind=Strong 2 1 Don't Play (0.5000 0.5000)
## 20) Outlook=Rainy 1 0 Don't Play (1.0000 0.0000) *
## 21) Outlook=Sunny 1 0 Play (0.0000 1.0000) *
## 11) Wind=Weak 3 0 Play (0.0000 1.0000) *
## 3) Outlook=Overcast 4 0 Play (0.0000 1.0000) *# Plot the Tree
rpart.plot(TennisTree,
shadow.col = "gray",
nn = TRUE,
main = "Classification Tree To Decide to Play Tennis")
2.5 Get Data to Determine Whether Customer Should be Classified as Credit Risk
# Input Data Set
RiskData <- data.frame(
Customer = c(1, 2, 3, 4, 5, 6, 7, 8),
Savings = c("Medium", "Low", "High", "Medium", "Low", "High", "Low", "Medium"),
Assets = c("High", "Low", "Medium", "Medium", "Medium", "High", "Low", "Medium"),
IncomeK = c(75, 50, 25, 50, 100, 25, 25, 75),
Risk = c("Good", "Bad", "Bad", "Good","Good", "Good", "Bad", "Good")
)
# Display Data Table
RiskData## Customer Savings Assets IncomeK Risk
## 1 1 Medium High 75 Good
## 2 2 Low Low 50 Bad
## 3 3 High Medium 25 Bad
## 4 4 Medium Medium 50 Good
## 5 5 Low Medium 100 Good
## 6 6 High High 25 Good
## 7 7 Low Low 25 Bad
## 8 8 Medium Medium 75 Good# Load Library if not available
if(! "rpart" %in% installed.packages()) { install.packages("rpart", dependencies = TRUE) }
library(rpart)
# Load Library if not available
if(! "vtable" %in% installed.packages()) { install.packages("vtable", dependencies = TRUE) }
library(vtable)
# Explore Data
cat("\n\nDescriptive Statistics of Columns in Data Frame:\n")##
##
## Descriptive Statistics of Columns in Data Frame:sumtable(RiskData, add.median = TRUE, out = "csv", simple.kable = TRUE, col.align = "right", align = "right", digits = 5,
title='Summary Statistics',
summ = list(
c('notNA(x)','mean(x)','sd(x)','min(x)', 'pctile(x)[25]', 'median(x)', 'pctile(x)[75]', 'max(x)', 'propNA(x)', 'getmode(x)'),
c('notNA(x)','mean(x)')
),
summ.names = list(
c('N','Mean','SD','Min','P25','P50','P75', 'Max','NA','Mode'),
c('Count','Percent')
)
)## Variable N Mean SD Min P25 P50 P75 Max NA Mode
## 1 Customer 8 4.5 2.4495 1 2.75 4.5 6.25 8 0
## 2 Savings 8
## 3 ... High 2 25%
## 4 ... Low 3 37.5%
## 5 ... Medium 3 37.5%
## 6 Assets 8
## 7 ... High 2 25%
## 8 ... Low 2 25%
## 9 ... Medium 4 50%
## 10 IncomeK 8 53.125 28.15 25 25 50 75 100 0
## 11 Risk 8
## 12 ... Bad 3 37.5%
## 13 ... Good 5 62.5%2.6 Use Predictors to Determine Whether Customer Should be Classified as Credit Risk
# Calculate the Tree
RiskTree <- rpart(
Risk ~ Savings + Assets + IncomeK,
data = RiskData,
method = "class",
minsplit = 2,
minbucket = 1
)
print(RiskTree)## n= 8
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 8 3 Good (0.3750 0.6250)
## 2) Assets=Low 2 0 Bad (1.0000 0.0000) *
## 3) Assets=High,Medium 6 1 Good (0.1667 0.8333)
## 6) Savings=High 2 1 Bad (0.5000 0.5000)
## 12) Assets=Medium 1 0 Bad (1.0000 0.0000) *
## 13) Assets=High 1 0 Good (0.0000 1.0000) *
## 7) Savings=Low,Medium 4 0 Good (0.0000 1.0000) *# Plot the Tree
rpart.plot(RiskTree,
shadow.col = "gray",
nn = TRUE,
main = "Classification Tree To Identify Credit Risk")
3 Classification and Regression Tree Using Conditional Inference Tree (CTREE, C4.5 Method)
3.1 Calculate Decision Tree Using Partykit
# Load Library if not available
if(! "partykit" %in% installed.packages()) { install.packages("partykit", dependencies = TRUE) }
library(partykit)# Calculate IncomeTree
TennisTree <- ctree(PlayTennis ~ Outlook + Humidity + Wind + Temperature, data = PlayTennis,
minbucket = 0,
minsplit = 1,
testtype = "Teststatistic",
mincriterion = 0,
weights = NULL)
TennisTree##
## Model formula:
## PlayTennis ~ Outlook + Humidity + Wind + Temperature
##
## Fitted party:
## [1] root
## | [2] Outlook in Overcast: Yes (n = 4, err = 0%)
## | [3] Outlook in Rainy, Sunny
## | | [4] Humidity in High
## | | | [5] Outlook in Rainy
## | | | | [6] Wind in Strong: No (n = 1, err = 0%)
## | | | | [7] Wind in Weak: Yes (n = 1, err = 0%)
## | | | [8] Outlook in Sunny: No (n = 3, err = 0%)
## | | [9] Humidity in Normal
## | | | [10] Wind in Strong
## | | | | [11] Temperature in Cool: No (n = 1, err = 0%)
## | | | | [12] Temperature in Mild: Yes (n = 1, err = 0%)
## | | | [13] Wind in Weak: Yes (n = 3, err = 0%)
##
## Number of inner nodes: 6
## Number of terminal nodes: 73.2 Plot Decision Tree to Play Tennis Using Partykit
# Load Library if not available
if(! "ggparty" %in% installed.packages()) { install.packages("ggparty", dependencies = TRUE) }
library(ggparty)
# Plot the Decsion Tree
columncol <- hcl(c(270, 260, 250), 200, 30, 0.6)
labelcol <- hcl(200, 200, 50, 0.2)
indexcol <- hcl(150, 200, 50, 0.4)
plot(TennisTree, main="Classification Tree: Are the Conditions Right to Play Tennis today?", cex.main = 1.5,
gp = gpar(fontsize = 11),
drop_terminal = TRUE, tnex = 1,
inner_panel = node_inner(TennisTree, abbreviate = FALSE,
fill = "lightgrey",
gp = gpar(),
pval = TRUE,
id = TRUE),
terminal_panel = node_barplot(TennisTree, col = "black",
fill = columncol[c(2,1,4)],
beside = TRUE,
ymax = 1,
ylines = TRUE,
widths = 1,
gap = 0.1,
reverse = FALSE,
id = TRUE))
3.3 Show Importance of Features
# Load Library if not available
if(! "vip" %in% installed.packages()) { install.packages("vip", dependencies = TRUE) }
library(vip)
# Create a variable importance plot
var_importance <- vip::vip(TennisTree, num_features = 4)
print(var_importance)
4 Classification and Regression Tree Using CART
4.1 Get Data for Predicting Whether Income is < 50k
# Download Data from URL4.2 Prepare Data
# Collapse Categrories
levels(Adults$marital.status)[2:4] <- "Married"
levels(Adults$workclass)[c(2,3,8)] <- "Gov"
levels(Adults$workclass)[c(5, 6)] <- "Self"
# Load Library if not available
if(! "stringr" %in% installed.packages()) { install.packages("stringr", dependencies = TRUE) }
library(stringr)
# Prepare Columns
names(Adults) <- str_to_title(names(Adults))
names(Adults) <- substr(names(Adults), 1, 6)
colnames(Adults)[2] <- "Class"
colnames(Adults)[5] <- "EduNum"
colnames(Adults)[6] <- "Marital"
colnames(Adults)[7] <- "Occupat"
colnames(Adults)[8] <- "Relationship"
colnames(Adults)[11] <- "CapGain"
colnames(Adults)[12] <- "CapLoss"
colnames(Adults)[13] <- "WeekHours"
head(Adults, 4)## Age Class Fnlwgt Educat EduNum Marital Occupat
## 1 25 Private 226802 11th 7 Never-married Machine-op-inspct
## 2 38 Private 89814 HS-grad 9 Married Farming-fishing
## 3 28 Gov 336951 Assoc-acdm 12 Married Protective-serv
## 4 44 Private 160323 Some-college 10 Married Machine-op-inspct
## Relationship Race Gender CapGain CapLoss WeekHours Native Income
## 1 Own-child Black Male 0 0 40 United-States <=50K
## 2 Husband White Male 0 0 50 United-States <=50K
## 3 Husband White Male 0 0 40 United-States >50K
## 4 Husband Black Male 7688 0 40 United-States >50K# Write data to working directory
write.csv(Adults, file = "Adults.csv") 4.3 Explore Data Frame
# Show Characteristics of Data Frame
cat("\n\nColumns Available in Data Frame:\n")##
##
## Columns Available in Data Frame:names(Adults)## [1] "Age" "Class" "Fnlwgt" "Educat" "EduNum"
## [6] "Marital" "Occupat" "Relationship" "Race" "Gender"
## [11] "CapGain" "CapLoss" "WeekHours" "Native" "Income"cat("\n\nShow Structure of the Data Frame:\n")##
##
## Show Structure of the Data Frame:str(Adults)## 'data.frame': 48842 obs. of 15 variables:
## $ Age : int 25 38 28 44 18 34 29 63 24 55 ...
## $ Class : Factor w/ 6 levels "?","Gov","Never-worked",..: 4 4 2 4 1 4 1 5 4 4 ...
## $ Fnlwgt : int 226802 89814 336951 160323 103497 198693 227026 104626 369667 104996 ...
## $ Educat : Factor w/ 16 levels "10th","11th",..: 2 12 8 16 16 1 12 15 16 6 ...
## $ EduNum : int 7 9 12 10 10 6 9 15 10 4 ...
## $ Marital : Factor w/ 5 levels "Divorced","Married",..: 3 2 2 2 3 3 3 2 3 2 ...
## $ Occupat : Factor w/ 15 levels "?","Adm-clerical",..: 8 6 12 8 1 9 1 11 9 4 ...
## $ Relationship: Factor w/ 6 levels "Husband","Not-in-family",..: 4 1 1 1 4 2 5 1 5 1 ...
## $ Race : Factor w/ 5 levels "Amer-Indian-Eskimo",..: 3 5 5 3 5 5 3 5 5 5 ...
## $ Gender : Factor w/ 2 levels "Female","Male": 2 2 2 2 1 2 2 2 1 2 ...
## $ CapGain : int 0 0 0 7688 0 0 0 3103 0 0 ...
## $ CapLoss : int 0 0 0 0 0 0 0 0 0 0 ...
## $ WeekHours : int 40 50 40 40 30 30 40 32 40 10 ...
## $ Native : Factor w/ 42 levels "?","Cambodia",..: 40 40 40 40 40 40 40 40 40 40 ...
## $ Income : Factor w/ 2 levels "<=50K",">50K": 1 1 2 2 1 1 1 2 1 1 ...cat("\n\nFirst 5 Rows of Data Frame:\n")##
##
## First 5 Rows of Data Frame:head(Adults, 5)## Age Class Fnlwgt Educat EduNum Marital Occupat
## 1 25 Private 226802 11th 7 Never-married Machine-op-inspct
## 2 38 Private 89814 HS-grad 9 Married Farming-fishing
## 3 28 Gov 336951 Assoc-acdm 12 Married Protective-serv
## 4 44 Private 160323 Some-college 10 Married Machine-op-inspct
## 5 18 ? 103497 Some-college 10 Never-married ?
## Relationship Race Gender CapGain CapLoss WeekHours Native Income
## 1 Own-child Black Male 0 0 40 United-States <=50K
## 2 Husband White Male 0 0 50 United-States <=50K
## 3 Husband White Male 0 0 40 United-States >50K
## 4 Husband Black Male 7688 0 40 United-States >50K
## 5 Own-child White Female 0 0 30 United-States <=50Kcat("\n\nDescriptive Statistics of Columns in Data Frame:\n")##
##
## Descriptive Statistics of Columns in Data Frame:st(Adults, add.median = TRUE, out = "csv", simple.kable = TRUE, col.align = "right", align = "right", digits = 5,
title='Summary Statistics',
summ = list(
c('notNA(x)','mean(x)','sd(x)','min(x)', 'pctile(x)[25]', 'median(x)', 'pctile(x)[75]', 'max(x)', 'propNA(x)', 'getmode(x)'),
c('notNA(x)','mean(x)')
),
summ.names = list(
c('N','Mean','SD','Min','P25','P50','P75', 'Max','NA','Mode'),
c('Count','Percent')
)
)## Variable N Mean SD Min P25 P50
## 1 Age 48842 38.644 13.711 17 28 37
## 2 Class 48842
## 3 ... ? 2799 5.731%
## 4 ... Gov 6549 13.409%
## 5 ... Never-worked 10 0.02%
## 6 ... Private 33906 69.42%
## 7 ... Self 5557 11.378%
## 8 ... Without-pay 21 0.043%
## 9 Fnlwgt 48842 189664 105604 12285 117551 178145
## 10 Educat 48842
## 11 ... 10th 1389 2.844%
## 12 ... 11th 1812 3.71%
## 13 ... 12th 657 1.345%
## 14 ... 1st-4th 247 0.506%
## 15 ... 5th-6th 509 1.042%
## 16 ... 7th-8th 955 1.955%
## 17 ... 9th 756 1.548%
## 18 ... Assoc-acdm 1601 3.278%
## 19 ... Assoc-voc 2061 4.22%
## 20 ... Bachelors 8025 16.431%
## 21 ... Doctorate 594 1.216%
## 22 ... HS-grad 15784 32.316%
## 23 ... Masters 2657 5.44%
## 24 ... Preschool 83 0.17%
## 25 ... Prof-school 834 1.708%
## 26 ... Some-college 10878 22.272%
## 27 EduNum 48842 10.078 2.571 1 9 10
## 28 Marital 48842
## 29 ... Divorced 6633 13.581%
## 30 ... Married 23044 47.181%
## 31 ... Never-married 16117 32.998%
## 32 ... Separated 1530 3.133%
## 33 ... Widowed 1518 3.108%
## 34 Occupat 48842
## 35 ... ? 2809 5.751%
## 36 ... Adm-clerical 5611 11.488%
## 37 ... Armed-Forces 15 0.031%
## 38 ... Craft-repair 6112 12.514%
## 39 ... Exec-managerial 6086 12.461%
## 40 ... Farming-fishing 1490 3.051%
## 41 ... Handlers-cleaners 2072 4.242%
## 42 ... Machine-op-inspct 3022 6.187%
## 43 ... Other-service 4923 10.079%
## 44 ... Priv-house-serv 242 0.495%
## 45 ... Prof-specialty 6172 12.637%
## 46 ... Protective-serv 983 2.013%
## 47 ... Sales 5504 11.269%
## 48 ... Tech-support 1446 2.961%
## 49 ... Transport-moving 2355 4.822%
## 50 Relationship 48842
## 51 ... Husband 19716 40.367%
## 52 ... Not-in-family 12583 25.763%
## 53 ... Other-relative 1506 3.083%
## 54 ... Own-child 7581 15.521%
## 55 ... Unmarried 5125 10.493%
## 56 ... Wife 2331 4.773%
## 57 Race 48842
## 58 ... Amer-Indian-Eskimo 470 0.962%
## 59 ... Asian-Pac-Islander 1519 3.11%
## 60 ... Black 4685 9.592%
## 61 ... Other 406 0.831%
## 62 ... White 41762 85.504%
## 63 Gender 48842
## 64 ... Female 16192 33.152%
## 65 ... Male 32650 66.848%
## 66 CapGain 48842 1079.1 7452 0 0 0
## 67 CapLoss 48842 87.502 403 0 0 0
## 68 WeekHours 48842 40.422 12.391 1 40 40
## 69 Native 48842
## 70 ... ? 857 1.755%
## 71 ... Cambodia 28 0.057%
## 72 ... Canada 182 0.373%
## 73 ... China 122 0.25%
## 74 ... Columbia 85 0.174%
## 75 ... Cuba 138 0.283%
## 76 ... Dominican-Republic 103 0.211%
## 77 ... Ecuador 45 0.092%
## 78 ... El-Salvador 155 0.317%
## 79 ... England 127 0.26%
## 80 ... France 38 0.078%
## 81 ... Germany 206 0.422%
## 82 ... Greece 49 0.1%
## 83 ... Guatemala 88 0.18%
## 84 ... Haiti 75 0.154%
## 85 ... Holand-Netherlands 1 0.002%
## 86 ... Honduras 20 0.041%
## 87 ... Hong 30 0.061%
## 88 ... Hungary 19 0.039%
## 89 ... India 151 0.309%
## 90 ... Iran 59 0.121%
## 91 ... Ireland 37 0.076%
## 92 ... Italy 105 0.215%
## 93 ... Jamaica 106 0.217%
## 94 ... Japan 92 0.188%
## 95 ... Laos 23 0.047%
## 96 ... Mexico 951 1.947%
## 97 ... Nicaragua 49 0.1%
## 98 ... Outlying-US(Guam-USVI-etc) 23 0.047%
## 99 ... Peru 46 0.094%
## 100 ... Philippines 295 0.604%
## 101 ... Poland 87 0.178%
## 102 ... Portugal 67 0.137%
## 103 ... Puerto-Rico 184 0.377%
## 104 ... Scotland 21 0.043%
## 105 ... South 115 0.235%
## 106 ... Taiwan 65 0.133%
## 107 ... Thailand 30 0.061%
## 108 ... Trinadad&Tobago 27 0.055%
## 109 ... United-States 43832 89.742%
## 110 ... Vietnam 86 0.176%
## 111 ... Yugoslavia 23 0.047%
## 112 Income 48842
## 113 ... <=50K 37155 76.072%
## 114 ... >50K 11687 23.928%
## P75 Max NA Mode
## 1 48 90 0
## 2
## 3
## 4
## 5
## 6
## 7
## 8
## 9 237642 1490400 0
## 10
## 11
## 12
## 13
## 14
## 15
## 16
## 17
## 18
## 19
## 20
## 21
## 22
## 23
## 24
## 25
## 26
## 27 12 16 0
## 28
## 29
## 30
## 31
## 32
## 33
## 34
## 35
## 36
## 37
## 38
## 39
## 40
## 41
## 42
## 43
## 44
## 45
## 46
## 47
## 48
## 49
## 50
## 51
## 52
## 53
## 54
## 55
## 56
## 57
## 58
## 59
## 60
## 61
## 62
## 63
## 64
## 65
## 66 0 99999 0
## 67 0 4356 0
## 68 45 99 0
## 69
## 70
## 71
## 72
## 73
## 74
## 75
## 76
## 77
## 78
## 79
## 80
## 81
## 82
## 83
## 84
## 85
## 86
## 87
## 88
## 89
## 90
## 91
## 92
## 93
## 94
## 95
## 96
## 97
## 98
## 99
## 100
## 101
## 102
## 103
## 104
## 105
## 106
## 107
## 108
## 109
## 110
## 111
## 112
## 113
## 1144.4 Standardise Numeric Variables
# Standardise Numeric Variables
Adults$Age.z <- (Adults$Age - mean(Adults$Age))/sd(Adults$Age)
Adults$EduNum.z <- (Adults$EduNum - mean(Adults$EduNum))/sd(Adults$EduNum)
Adults$CapGain.z <- (Adults$CapGain - mean(Adults$CapGain))/sd(Adults$CapGain)
Adults$CapLoss.z <- (Adults$CapLoss - mean(Adults$CapLoss))/sd(Adults$CapLoss)
Adults$WeekHours.z <- (Adults$WeekHours - mean(Adults$WeekHours))/sd(Adults$WeekHours)
names(Adults)## [1] "Age" "Class" "Fnlwgt" "Educat" "EduNum"
## [6] "Marital" "Occupat" "Relationship" "Race" "Gender"
## [11] "CapGain" "CapLoss" "WeekHours" "Native" "Income"
## [16] "Age.z" "EduNum.z" "CapGain.z" "CapLoss.z" "WeekHours.z"4.5 Use Predictors to Classify Whether or not a Person’s Income is less than $50K
# Calculate the Income Tree
IncomeTree <- rpart(Income ~ Age.z + EduNum.z + CapGain.z + CapLoss.z + WeekHours.z + Race + Gender + Class + Marital, data = Adults, method = "class", minsplit = 2, minbucket = 1)
print(IncomeTree)## n= 48842
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 48842 11690 <=50K (0.76072 0.23928)
## 2) Marital=Divorced,Never-married,Separated,Widowed 25798 1631 <=50K (0.93678 0.06322)
## 4) CapGain.z< 0.802 25352 1200 <=50K (0.95267 0.04733) *
## 5) CapGain.z>=0.802 446 15 >50K (0.03363 0.96637) *
## 3) Marital=Married 23044 10060 <=50K (0.56362 0.43638)
## 6) EduNum.z< 0.942 16234 5229 <=50K (0.67790 0.32210)
## 12) CapGain.z< 0.539 15465 4475 <=50K (0.71064 0.28936)
## 24) EduNum.z< -0.6138 2672 266 <=50K (0.90045 0.09955) *
## 25) EduNum.z>=-0.6138 12793 4209 <=50K (0.67099 0.32901)
## 50) CapLoss.z< 4.3 12298 3833 <=50K (0.68832 0.31168) *
## 51) CapLoss.z>=4.3 495 119 >50K (0.24040 0.75960) *
## 13) CapGain.z>=0.539 769 15 >50K (0.01951 0.98049) *
## 7) EduNum.z>=0.942 6810 1983 >50K (0.29119 0.70881) *4.6 Plot Decision Tree using Cart
# Plot the Decsion Tree
cat("\n\nThe following information is shown in each node:\n")##
##
## The following information is shown in each node:cat("- Node Number\n")## - Node Numbercat("- Predicted class\n")## - Predicted classcat("- Predicted probability of this class\n")## - Predicted probability of this classcat("- Percentage of Observations in the node\n")## - Percentage of Observations in the noderpart.plot(IncomeTree, main = "Classification Tree: Is a Person Likely to Earn 50k?",
under = FALSE,
type = 2,
extra = 106,
clip.right.labs = TRUE, shadow.col = "gray", # shadows under the node boxes
nn = TRUE,
fallen.leaves = TRUE,
digits = 3,
box.palette = "RdGn"
)
5 Classification and Regression Tree Using Conditional Inference Tree (CTREE, C4.5 Method)
5.1 Calculate Decision Tree Using Partykit and CTREE
# Load Library if not available
if(! "partykit" %in% installed.packages()) { install.packages("partykit", dependencies = TRUE) }
library(partykit)
# Load Library if not available
if(! "ggparty" %in% installed.packages()) { install.packages("ggparty", dependencies = TRUE) }
library(ggparty)
# Calculate IncomeTree
IncomeTree <- ctree(Income ~ Age + EduNum + CapGain + CapLoss + WeekHours + Race + Gender + Class + Marital, data = Adults, weights = NULL,
minsplit = 12,
maxdepth = 3,
teststat = "quadratic",
testtype = "Bonferroni",
minbucket = 10,
mincriterion = 0.95,)
# Calculate IncomeTree from Normalised Values
IncomeTree.z <- ctree(Income ~ Age.z + EduNum.z + CapGain.z + CapLoss.z + WeekHours.z + Race + Gender + Class + Marital, data = Adults, weights = NULL,
minsplit = 12,
maxdepth = 3,
teststat = "quadratic",
testtype = "Bonferroni",
minbucket = 10,
mincriterion = 0.95,)
print(IncomeTree)##
## Model formula:
## Income ~ Age + EduNum + CapGain + CapLoss + WeekHours + Race +
## Gender + Class + Marital
##
## Fitted party:
## [1] root
## | [2] Marital in Divorced, Never-married, Separated, Widowed
## | | [3] CapGain <= 6849
## | | | [4] EduNum <= 13: <=50K (n = 23975, err = 4%)
## | | | [5] EduNum > 13: <=50K (n = 1377, err = 25%)
## | | [6] CapGain > 6849
## | | | [7] WeekHours <= 35: >50K (n = 46, err = 20%)
## | | | [8] WeekHours > 35: >50K (n = 400, err = 2%)
## | [9] Marital in Married
## | | [10] EduNum <= 12
## | | | [11] EduNum <= 8: <=50K (n = 2726, err = 12%)
## | | | [12] EduNum > 8: <=50K (n = 13508, err = 36%)
## | | [13] EduNum > 12
## | | | [14] WeekHours <= 31: <=50K (n = 567, err = 44%)
## | | | [15] WeekHours > 31: >50K (n = 6243, err = 27%)
##
## Number of inner nodes: 7
## Number of terminal nodes: 8print(IncomeTree.z)##
## Model formula:
## Income ~ Age.z + EduNum.z + CapGain.z + CapLoss.z + WeekHours.z +
## Race + Gender + Class + Marital
##
## Fitted party:
## [1] root
## | [2] Marital in Divorced, Never-married, Separated, Widowed
## | | [3] CapGain.z <= 0.77
## | | | [4] EduNum.z <= 1.14: <=50K (n = 23975, err = 4%)
## | | | [5] EduNum.z > 1.14: <=50K (n = 1377, err = 25%)
## | | [6] CapGain.z > 0.77
## | | | [7] WeekHours.z <= -0.44: >50K (n = 46, err = 20%)
## | | | [8] WeekHours.z > -0.44: >50K (n = 400, err = 2%)
## | [9] Marital in Married
## | | [10] EduNum.z <= 0.75
## | | | [11] EduNum.z <= -0.81: <=50K (n = 2726, err = 12%)
## | | | [12] EduNum.z > -0.81: <=50K (n = 13508, err = 36%)
## | | [13] EduNum.z > 0.75
## | | | [14] WeekHours.z <= -0.76: <=50K (n = 567, err = 44%)
## | | | [15] WeekHours.z > -0.76: >50K (n = 6243, err = 27%)
##
## Number of inner nodes: 7
## Number of terminal nodes: 85.2 Plot Decision Tree Using Partykit
# Plot the Decsion Tree
columncol <- hcl(c(270, 260, 250), 200, 30, 0.6)
labelcol <- hcl(200, 200, 50, 0.2)
indexcol <- hcl(150, 200, 50, 0.4)
p <- plot(IncomeTree, main="Classification Tree: Is a Person Likely to Earn 50k?", cex.main = 1.5,
gp = gpar(fontsize = 11),
drop_terminal = TRUE, tnex = 1,
inner_panel = node_inner(IncomeTree, abbreviate = FALSE,
fill = "lightgrey",
gp = gpar(),
pval = TRUE,
id = TRUE),
terminal_panel = node_barplot(IncomeTree, col = "black",
fill = columncol[c(2,1,4)],
beside = TRUE,
ymax = 1,
ylines = TRUE,
widths = 1,
gap = 0.1,
reverse = FALSE,
id = TRUE))
# Save the plot
ggsave(filename = paste("Is a Person Likely to Earn 50k", ".png", sep = ""), plot = p, width = 12, height = 8) 6 Classification and Regression Tree Using Conditional Inference Tree (C5.0, C5.0 Method)
6.1 Calculate Decision Tree Using C5.0
# Load Library if not available
if(! "C50" %in% installed.packages()) { install.packages("C50", dependencies = TRUE) }
library(C50)
# Make Variables
Outcome <- as.factor(Adults$Income)
Factors <- data.frame(Adults[, c(1, 2, 4, 6, 9, 10, 13, 14)])
# Calculate IncomeTree from Observed Values
IncomeTree <- C5.0(y = Outcome, x = Factors, rules = FALSE, control = C5.0Control(
subset = FALSE,
bands = 0,
winnow = FALSE,
noGlobalPruning = FALSE,
CF = 0.05,
minCases = 50,
fuzzyThreshold = FALSE,
seed = sample.int(4096, size = 1) - 1L,
earlyStopping = TRUE,
label = "outcome"))
# Calculate IncomeTree from Normalised Values
#IncomeTree.z <- C5.0(Income ~ Age.z + EduNum.z + CapGain.z + CapLoss.z + WeekHours.z + Race + Gender + Class + Marital, data = Adults, rules = FALSE)
summary(IncomeTree)##
## Call:
## C5.0.default(x = Factors, y = Outcome, rules = FALSE, control
## = FALSE, CF = 0.05, minCases = 50, fuzzyThreshold = FALSE, seed
## = sample.int(4096, size = 1) - 1L, earlyStopping = TRUE, label = "outcome"))
##
##
## C5.0 [Release 2.07 GPL Edition] Wed Jan 22 22:08:14 2025
## -------------------------------
##
## Class specified by attribute `outcome'
##
## Read 48842 cases (9 attributes) from undefined.data
##
## Decision tree:
##
## Age <= 27: <=50K (12012/369)
## Age > 27:
## :...Marital in {Divorced,Never-married,Separated,
## : Widowed}: <=50K (15612/1529)
## Marital = Married:
## :...Educat in {10th,11th,12th,1st-4th,5th-6th,7th-8th,9th,Assoc-voc,
## : HS-grad,Preschool}: <=50K (10153/2939)
## Educat in {Doctorate,Masters,Prof-school}: >50K (2539/530)
## Educat = Assoc-acdm:
## :...Age <= 35: <=50K (180/68)
## : Age > 35: >50K (482/206)
## Educat = Bachelors:
## :...WeekHours <= 31: <=50K (323/135)
## : WeekHours > 31: >50K (3682/1094)
## Educat = Some-college:
## :...WeekHours <= 34: <=50K (353/92)
## WeekHours > 34:
## :...Class in {?,Never-worked}: >50K (0)
## Class in {Self,Without-pay}: <=50K (615.9/271)
## Class = Gov:
## :...Age <= 38: <=50K (185/71.6)
## : Age > 38: >50K (364.7/157.7)
## Class = Private:
## :...WeekHours > 41:
## :...Age <= 37: <=50K (348.3/139)
## : Age > 37: >50K (608/235.3)
## WeekHours <= 41:
## :...Age > 61: <=50K (65/25)
## Age <= 61:
## :...Age <= 47: <=50K (982.7/403)
## Age > 47: >50K (336.3/143.7)
##
##
## Evaluation on training data (48842 cases):
##
## Decision Tree
## ----------------
## Size Errors
##
## 17 8408(17.2%) <<
##
##
## (a) (b) <-classified as
## ---- ----
## 34787 2368 (a): class <=50K
## 6040 5647 (b): class >50K
##
##
## Attribute usage:
##
## 100.00% Age
## 75.41% Marital
## 43.44% Educat
## 16.10% WeekHours
## 7.04% Class
##
##
## Time: 0.1 secs#summary(IncomeTree.z)6.2 Plot Decision Tree Using C5.0
# Plot the Decsion Tree
columncol <- hcl(c(270, 260, 250), 200, 30, 0.6)
labelcol <- hcl(200, 200, 50, 0.2)
indexcol <- hcl(150, 200, 50, 0.4)
plot(IncomeTree, main="Classification Tree: Is a Person Likely to Earn 50k or more?", cex.main = 1.5,
gp = gpar(fontsize = 15),
drop_terminal = FALSE, tnex = 1,
)
6.3 Get Data to Determine Whether Customer Should be Classified as Credit Risk
# Input Data Set
RiskData <- data.frame(
Customer = c(1, 2, 3, 4, 5, 6, 7, 8),
Savings = c("Medium", "Low", "High", "Medium", "Low", "High", "Low", "Medium"),
Assets = c("High", "Low", "Medium", "Medium", "Medium", "High", "Low", "Medium"),
IncomeK = c(75, 50, 25, 50, 100, 25, 25, 75),
Risk = c("Good", "Bad", "Bad", "Good","Good", "Good", "Bad", "Good")
)
# Display Data Table
RiskData## Customer Savings Assets IncomeK Risk
## 1 1 Medium High 75 Good
## 2 2 Low Low 50 Bad
## 3 3 High Medium 25 Bad
## 4 4 Medium Medium 50 Good
## 5 5 Low Medium 100 Good
## 6 6 High High 25 Good
## 7 7 Low Low 25 Bad
## 8 8 Medium Medium 75 Good# Explore Data
cat("\n\nDescriptive Statistics of Columns in Data Frame:\n")##
##
## Descriptive Statistics of Columns in Data Frame:sumtable(RiskData, add.median = TRUE, out = "csv", simple.kable = TRUE, col.align = "right", align = "right", digits = 5,
title='Summary Statistics',
summ = list(
c('notNA(x)','mean(x)','sd(x)','min(x)', 'pctile(x)[25]', 'median(x)', 'pctile(x)[75]', 'max(x)', 'propNA(x)', 'getmode(x)'),
c('notNA(x)','mean(x)')
),
summ.names = list(
c('N','Mean','SD','Min','P25','P50','P75', 'Max','NA','Mode'),
c('Count','Percent')
)
)## Variable N Mean SD Min P25 P50 P75 Max NA Mode
## 1 Customer 8 4.5 2.4495 1 2.75 4.5 6.25 8 0
## 2 Savings 8
## 3 ... High 2 25%
## 4 ... Low 3 37.5%
## 5 ... Medium 3 37.5%
## 6 Assets 8
## 7 ... High 2 25%
## 8 ... Low 2 25%
## 9 ... Medium 4 50%
## 10 IncomeK 8 53.125 28.15 25 25 50 75 100 0
## 11 Risk 8
## 12 ... Bad 3 37.5%
## 13 ... Good 5 62.5%# Write data to working directory
write.csv(RiskData, file = "RiskData.csv")6.4 Use Predictors to Determine Whether Customer Should be Classified as Credit Risk Using C5.0
# Make Variables
Outcome <- as.factor(RiskData$Risk)
Factors <- data.frame(RiskData[, c(2, 3, 4)])
# Calculate IncomeTree from Observed Values
RiskTree <- C5.0(Outcome ~ Savings + Assets + IncomeK, data = RiskData, rules = FALSE)
summary(RiskTree)##
## Call:
## C5.0.formula(formula = Outcome ~ Savings + Assets + IncomeK, data =
## RiskData, rules = FALSE)
##
##
## C5.0 [Release 2.07 GPL Edition] Wed Jan 22 22:08:15 2025
## -------------------------------
##
## Class specified by attribute `outcome'
##
## Read 8 cases (4 attributes) from undefined.data
##
## Decision tree:
##
## Assets in {High,Medium}: Good (6/1)
## Assets = Low: Bad (2)
##
##
## Evaluation on training data (8 cases):
##
## Decision Tree
## ----------------
## Size Errors
##
## 2 1(12.5%) <<
##
##
## (a) (b) <-classified as
## ---- ----
## 2 1 (a): class Bad
## 5 (b): class Good
##
##
## Attribute usage:
##
## 100.00% Assets
##
##
## Time: 0.0 secs6.5 Plot Decision Tree to Determine Whether Customer Should be Classified as Credit Risk Using C5.0
# Plot the Decision Tree
columncol <- hcl(c(270, 260, 250), 200, 30, 0.6)
labelcol <- hcl(200, 200, 50, 0.2)
indexcol <- hcl(150, 200, 50, 0.4)
plot(RiskTree, main="Classification Tree: Is a Person Likely to be Credit Risk?", cex.main = 1.5,
gp = gpar(fontsize = 11),
drop_terminal = TRUE, tnex = 1,
)
7 Assignment
7.1 Get Data to Classify Whether or not a Car Insurance Claim was Fraudulent
# Load Library if not available
if(! "rpart" %in% installed.packages()) { install.packages("rpart", dependencies = TRUE) }
library(rpart)
if(! "rpart.plot" %in% installed.packages()) { install.packages("rpart.plot", dependencies = TRUE) }
library(rpart.plot)
# Set number of decimals
options(digits = 4) # Modify global options7.2 Get Data for Predicting Whether Going For Morning Walk
# Create the dataset
Exercise <- data.frame(
Rain = c('No', 'Yes', 'Yes', 'No', 'No', 'No', 'No', 'No', 'Yes', 'Yes', 'Yes', 'Yes'),
EarlyMeeting = c('No', 'No', 'Yes', 'Yes', 'Yes', 'Yes', 'No', 'No', 'Yes', 'No', 'No', 'Yes'),
ExerciseEveningBefore = c('No', 'No', 'Yes', 'Yes', 'Yes', 'Yes', 'No', 'Yes', 'No', 'No', 'Yes', 'No'),
GoForWalk = c('Yes', 'No', 'No', 'No', 'No', 'Yes', 'Yes', 'Yes', 'No', 'No', 'No', 'No')
)## Day Rain EarlyMeeting ExerciseEveningBefore GoForWalk
## 1 1 No No No Yes
## 2 2 Yes No No No
## 3 3 Yes Yes Yes No
## 4 4 No Yes Yes No
## 5 5 No Yes Yes No
## 6 6 No Yes Yes No
## 7 7 No No No Yes
## 8 8 No No Yes Yes
## 9 9 Yes Yes No No
## 10 10 Yes No No No
## 11 11 Yes No Yes No
## 12 12 Yes Yes No No
## 13 13 No No Yes Yes
## 14 14 No No Yes Yes
## 15 15 No Yes No Yes
## 16 16 No Yes No Yes7.3 Calculate Decision Tree
# Calculate the Tree
control <- rpart.control(minsplit = 1, minbucket = 1, cp = 0.001, maxdepth = 30)
mytree <- rpart(
GoForWalk ~ Rain + EarlyMeeting + ExerciseEveningBefore, data = Exercise,
method = "class",
control = control
)
print(mytree)## n= 16
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 16 7 No (0.5625 0.4375)
## 2) Rain=Yes 6 0 No (1.0000 0.0000) *
## 3) Rain=No 10 3 Yes (0.3000 0.7000)
## 6) EarlyMeeting=Yes 5 2 No (0.6000 0.4000)
## 12) ExerciseEveningBefore=Yes 3 0 No (1.0000 0.0000) *
## 13) ExerciseEveningBefore=No 2 0 Yes (0.0000 1.0000) *
## 7) EarlyMeeting=No 5 0 Yes (0.0000 1.0000) *# Plot the Tree
cat("\n\nThe following information is shown in each node:\n")##
##
## The following information is shown in each node:cat("- Node number\n")## - Node numbercat("- Predicted class\n")## - Predicted classcat("- Predicted probability of this class\n")## - Predicted probability of this classcat("- Percentage of observations in the node\n")## - Percentage of observations in the noderpart.plot(mytree, main = "Classification Tree To Decide Going for Morning Walk",
under = FALSE,
type = 2,
extra = 106,
clip.right.labs = TRUE, shadow.col = "gray", # shadows under the node boxes
nn = TRUE,
fallen.leaves = TRUE,
digits = 3,
box.palette = "RdGn"
)
7.4 Calculate Decision Tree Using PartyKit
# Load Library if not available
if(! "partykit" %in% installed.packages()) { install.packages("partykit", dependencies = TRUE) }
library(partykit)
if(! "ggplot2" %in% installed.packages()) { install.packages("ggplot2", dependencies = TRUE) }
library(ggplot2)
if(! "ggparty" %in% installed.packages()) { install.packages("ggparty", dependencies = TRUE) }
library(ggparty)
# Calculate ExerciseTree
ExerciseTree <- ctree(GoForWalk ~ Rain + EarlyMeeting + ExerciseEveningBefore, data = Exercise,
minsplit = 1,
minbucket = 1)
ExerciseTree##
## Model formula:
## GoForWalk ~ Rain + EarlyMeeting + ExerciseEveningBefore
##
## Fitted party:
## [1] root
## | [2] Rain in No: Yes (n = 10, err = 30%)
## | [3] Rain in Yes: No (n = 6, err = 0%)
##
## Number of inner nodes: 1
## Number of terminal nodes: 28 Plot Decision Tree Using Partykit
# Plot the Decsion Tree
columncol <- hcl(c(270, 260, 250), 200, 30, 0.6)
labelcol <- hcl(200, 200, 50, 0.2)
indexcol <- hcl(150, 200, 50, 0.4)
plot(ExerciseTree, main="Classification Tree: Are we going For a Walk This Morning?", cex.main = 1.5,
gp = gpar(fontsize = 11),
drop_terminal = TRUE, tnex = 1,
inner_panel = node_inner(ExerciseTree, abbreviate = FALSE,
fill = "lightgrey",
gp = gpar(),
pval = TRUE,
id = TRUE),
terminal_panel = node_barplot(ExerciseTree, col = "black",
fill = columncol[c(2,1,4)],
beside = TRUE,
ymax = 1,
ylines = TRUE,
widths = 1,
gap = 0.1,
reverse = FALSE,
id = TRUE))
8.1 Calculate Decision Tree Using C5.0
# Load Library if not available
if(! "C50" %in% installed.packages()) { install.packages("C50", dependencies = TRUE) }
library(C50)
# Make Variables
Outcome <- as.factor(Exercise$GoForWalk)
Factors <- data.frame(
Exercise$Rain, Exercise$EarlyMeeting, Exercise$ExerciseEveningBefore)
# Calculate ExerciseTree from Observed Values
ExerciseTree <- C5.0(y = Outcome, x = Factors, rules = FALSE, trials = 5, control = C5.0Control(
subset = FALSE,
bands = 0,
winnow = FALSE,
noGlobalPruning = FALSE,
CF = 0.05,
minCases = 1,
fuzzyThreshold = FALSE,
seed = sample.int(4096, size = 1) - 1L,
earlyStopping = FALSE,
label = "outcome"))
summary(ExerciseTree)##
## Call:
## C5.0.default(x = Factors, y = Outcome, trials = 5, rules = FALSE, control
## = FALSE, CF = 0.05, minCases = 1, fuzzyThreshold = FALSE, seed
## = sample.int(4096, size = 1) - 1L, earlyStopping = FALSE, label = "outcome"))
##
##
## C5.0 [Release 2.07 GPL Edition] Wed Jan 22 22:08:17 2025
## -------------------------------
##
## Class specified by attribute `outcome'
##
## Read 16 cases (4 attributes) from undefined.data
##
## ----- Trial 0: -----
##
## Decision tree:
##
## Exercise.Rain = Yes: No (6)
## Exercise.Rain = No:
## :...Exercise.EarlyMeeting = No: Yes (5)
## Exercise.EarlyMeeting = Yes:
## :...Exercise.ExerciseEveningBefore = No: Yes (2)
## Exercise.ExerciseEveningBefore = Yes: No (3)
##
## ----- Trial 1: -----
##
## Decision tree:
##
## Exercise.Rain = No: Yes (7.5/2.2)
## Exercise.Rain = Yes: No (4.5)
##
## ----- Trial 2: -----
##
## Decision tree:
##
## Exercise.Rain = Yes: No (4.8)
## Exercise.Rain = No:
## :...Exercise.EarlyMeeting = No: Yes (4)
## Exercise.EarlyMeeting = Yes:
## :...Exercise.ExerciseEveningBefore = No: Yes (1.6)
## Exercise.ExerciseEveningBefore = Yes: No (5.5)
##
## ----- Trial 3: -----
##
## Decision tree:
## No (12/4.2)
##
## ----- Trial 4: -----
##
## Decision tree:
##
## Exercise.Rain = Yes: No (4.3)
## Exercise.Rain = No:
## :...Exercise.EarlyMeeting = No: Yes (4.9)
## Exercise.EarlyMeeting = Yes:
## :...Exercise.ExerciseEveningBefore = No: Yes (2)
## Exercise.ExerciseEveningBefore = Yes: No (4.9)
##
##
## Evaluation on training data (16 cases):
##
## Trial Decision Tree
## ----- ----------------
## Size Errors
##
## 0 4 0( 0.0%)
## 1 2 3(18.8%)
## 2 4 0( 0.0%)
## 3 1 7(43.8%)
## 4 4 0( 0.0%)
## boost 0( 0.0%) <<
##
##
## (a) (b) <-classified as
## ---- ----
## 9 (a): class No
## 7 (b): class Yes
##
##
## Attribute usage:
##
## 100.00% Exercise.Rain
## 62.50% Exercise.EarlyMeeting
## 31.25% Exercise.ExerciseEveningBefore
##
##
## Time: 0.0 secs8.2 Plot Decision Tree Using C5.0
# Plot the Decsion Tree
columncol <- hcl(c(270, 260, 250), 200, 30, 0.6)
labelcol <- hcl(200, 200, 50, 0.2)
indexcol <- hcl(150, 200, 50, 0.4)
plot(ExerciseTree, main="Classification Tree: Are we Going for a Walk this Morning?", cex.main = 1.5,
gp = gpar(fontsize = 15),
drop_terminal = FALSE, tnex = 1,
)
9 Classification Using Random Forest
9.1 Get Data and Library
# Load Library if not available
if(! "randomForest" %in% installed.packages()) { install.packages("randomForest", dependencies = TRUE) }
library(randomForest)
# Load the iris dataset
data(iris)
# View the first few rows of the dataset
head(iris)## Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## 1 5.1 3.5 1.4 0.2 setosa
## 2 4.9 3.0 1.4 0.2 setosa
## 3 4.7 3.2 1.3 0.2 setosa
## 4 4.6 3.1 1.5 0.2 setosa
## 5 5.0 3.6 1.4 0.2 setosa
## 6 5.4 3.9 1.7 0.4 setosa9.2 Split the Data
# Set seed for reproducibility
set.seed(123)
# Split the data into training (70%) and testing (30%) sets
train_index <- sample(1:nrow(iris), 0.7 * nrow(iris))
train_data <- iris[train_index, ]
test_data <- iris[-train_index, ]9.3 Train Random Forest Model
# Train the random forest model
rf_model <- randomForest(Species ~ ., data = train_data, importance = TRUE, ntree = 100)
# Print the model summary
print(rf_model)##
## Call:
## randomForest(formula = Species ~ ., data = train_data, importance = TRUE, ntree = 100)
## Type of random forest: classification
## Number of trees: 100
## No. of variables tried at each split: 2
##
## OOB estimate of error rate: 4.76%
## Confusion matrix:
## setosa versicolor virginica class.error
## setosa 36 0 0 0.00000
## versicolor 0 29 3 0.09375
## virginica 0 2 35 0.054059.4 Evaluate Model on Testing Data
# Make predictions on the testing set
predictions <- predict(rf_model, test_data)
# Create a confusion matrix
confusion_matrix <- table(predictions, test_data$Species)
# Print the confusion matrix
print(confusion_matrix)##
## predictions setosa versicolor virginica
## setosa 14 0 0
## versicolor 0 17 0
## virginica 0 1 13# Calculate accuracy
accuracy <- sum(diag(confusion_matrix)) / sum(confusion_matrix)
print(paste("Accuracy:", round(accuracy * 100, 2), "%"))## [1] "Accuracy: 97.78 %"9.5 Analyse Variable Importance
# View the importance of each variable
importance(rf_model)## setosa versicolor virginica MeanDecreaseAccuracy MeanDecreaseGini
## Sepal.Length 2.485 4.1775 1.214 4.456 5.829
## Sepal.Width 1.725 -0.2708 3.803 3.246 1.586
## Petal.Length 9.533 13.3983 13.888 15.158 29.960
## Petal.Width 10.340 12.5554 13.980 15.020 31.708# Plot variable importance
varImpPlot(rf_model)