Task:using a corpus of labeled spam and ham (non-spam) e-mails to predict whether or not a new document is spam. Train SVM model for classification and predict the test data.

load library

library(rvest)
library(httr)
library(tm)

## Loading required package: NLP

## 
## Attaching package: 'NLP'

## The following object is masked from 'package:httr':
## 
##     content

library(tidytext)
library(RTextTools)

## Loading required package: SparseM

## 
## Attaching package: 'SparseM'

## The following object is masked from 'package:base':
## 
##     backsolve

library(knitr)
library(tidyverse)

## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr     1.1.4     ✔ readr     2.1.5
## ✔ forcats   1.0.0     ✔ stringr   1.5.1
## ✔ ggplot2   3.4.4     ✔ tibble    3.2.1
## ✔ lubridate 1.9.3     ✔ tidyr     1.3.1
## ✔ purrr     1.0.2

## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ ggplot2::annotate()     masks NLP::annotate()
## ✖ NLP::content()          masks httr::content()
## ✖ dplyr::filter()         masks stats::filter()
## ✖ readr::guess_encoding() masks rvest::guess_encoding()
## ✖ dplyr::lag()            masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(wordcloud)

## Loading required package: RColorBrewer

library(e1071)

#load files from github desktop

Read ham and spam files into R ham=0 spam=1

ham <- "/Users/angelgallardo/Documents/GitHub/DATA607-P4/easy_ham"

hamfiles = list.files(ham)

hamlist <- NA
for(i in 1:length(hamfiles))
{
  filepath<-paste0(ham, "/", hamfiles[1])  
  text <-readLines(filepath)
  list1<- list(paste(text, collapse="\n"))
  hamlist = c(hamlist,list1)
  
}

# ham data frame
hamDF <-as.data.frame(unlist(hamlist),stringsAsFactors = FALSE)
hamDF$type <- "0"
colnames(hamDF) <- c("text","type")

# Create Spam Dataframe
spam <- "/Users/angelgallardo/Documents/GitHub/DATA607-P4/spam"
spamfiles = list.files(spam)

spamlist <- NA
for(i in 1:length(spamfiles))
{
  filepath<-paste0(spam, "/", spamfiles[1])  
  text <-readLines(filepath)
  list1<- list(paste(text, collapse="\n"))
  spamlist = c(spamlist,list1)
  
}

spamDF <-as.data.frame(unlist(spamlist),stringsAsFactors = FALSE)
spamDF$type <- "1"
colnames(spamDF) <- c("text","type")


# creating combined data frame of spam and ham
spam_ham_df <- rbind(hamDF, spamDF)

prepare corpus and Preprocessing - remove numbers, puncuations, sotpwords and whitespace

emails <- Corpus(VectorSource(spam_ham_df$text))
cleanCorpus <- tm_map(emails, removeNumbers)

## Warning in tm_map.SimpleCorpus(emails, removeNumbers): transformation drops
## documents

cleanCorpus <- tm_map(cleanCorpus, removePunctuation)

## Warning in tm_map.SimpleCorpus(cleanCorpus, removePunctuation): transformation
## drops documents

cleanCorpus <- tm_map(cleanCorpus, removeWords, stopwords())

## Warning in tm_map.SimpleCorpus(cleanCorpus, removeWords, stopwords()):
## transformation drops documents

cleanCorpus <- tm_map(cleanCorpus, stripWhitespace)

## Warning in tm_map.SimpleCorpus(cleanCorpus, stripWhitespace): transformation
## drops documents

sample 80% data training and 30 % for prediction

sample <- floor(0.80 * nrow(spam_ham_df))

# set the seed to make your partition reproductible
set.seed(555)
train_ind <- sample(seq_len(nrow(spam_ham_df)), size = sample)

train_spam_ham <- spam_ham_df[train_ind, ]
test_spam_ham <- spam_ham_df[-train_ind, ]

# count of spam and ham in train data set
spam<-subset(train_spam_ham,train_spam_ham$type == "0")
ham<-subset(train_spam_ham,train_spam_ham$type == "1")

Create corpus for training and test data

train_email_corpus <- Corpus(VectorSource(train_spam_ham$text))
test_email_corpus <- Corpus(VectorSource(test_spam_ham$text))

train_clean_corpus <- tm_map(train_email_corpus ,removeNumbers)

## Warning in tm_map.SimpleCorpus(train_email_corpus, removeNumbers):
## transformation drops documents

test_clean_corpus <- tm_map(test_email_corpus, removeNumbers)

## Warning in tm_map.SimpleCorpus(test_email_corpus, removeNumbers):
## transformation drops documents

train_clean_corpus <- tm_map(train_clean_corpus, removePunctuation)

## Warning in tm_map.SimpleCorpus(train_clean_corpus, removePunctuation):
## transformation drops documents

test_clean_corpus <- tm_map(test_clean_corpus, removePunctuation)

## Warning in tm_map.SimpleCorpus(test_clean_corpus, removePunctuation):
## transformation drops documents

train_clean_corpus <- tm_map(train_clean_corpus, removeWords, stopwords())

## Warning in tm_map.SimpleCorpus(train_clean_corpus, removeWords, stopwords()):
## transformation drops documents

test_clean_corpus  <- tm_map(test_clean_corpus, removeWords, stopwords())

## Warning in tm_map.SimpleCorpus(test_clean_corpus, removeWords, stopwords()):
## transformation drops documents

train_clean_corpus<- tm_map(train_clean_corpus, stripWhitespace)

## Warning in tm_map.SimpleCorpus(train_clean_corpus, stripWhitespace):
## transformation drops documents

test_clean_corpus<- tm_map(test_clean_corpus, stripWhitespace)

## Warning in tm_map.SimpleCorpus(test_clean_corpus, stripWhitespace):
## transformation drops documents

train_email_dtm <- DocumentTermMatrix(train_clean_corpus)
test_email_dtm <- DocumentTermMatrix(test_clean_corpus)

Train and predict using the SVM model 99% accuracy

train_matrix <- as.matrix(train_email_dtm)

# Extract labels from training data
labels <- train_spam_ham$type  

# Ensure labels are factors for classification
labels <- as.factor(labels)

# Train SVM model for classification
svm_model <- svm(x = train_matrix, y = labels, kernel = "linear", cost = 1)


# Convert Document-Term Matrix of test data to matrix
test_matrix <- as.matrix(test_email_dtm)

# Make predictions on test data using the trained SVM model
predictions <- predict(svm_model, newdata = test_matrix)

# Evaluate model performance (e.g., accuracy)
accuracy <- mean(predictions == test_spam_ham$type)
cat("Accuracy:", accuracy, "\n")

## Accuracy: 0.9983633

Test

#(ham) 515 instances correctly predicted as class 0 #(spam): 1 instance incorrectly predicted as class 0, # and 95 instances correctly predicted as class 1

# Create confusion matrix
conf_matrix <- table(Actual = test_spam_ham$type, Predicted = predictions)
print(conf_matrix)

##       Predicted
## Actual   0   1
##      0 515   0
##      1   1  95

# Calculate precision, recall, and F1-score
precision <- conf_matrix[2, 2] / sum(conf_matrix[, 2])
recall <- conf_matrix[2, 2] / sum(conf_matrix[2, ])
f1_score <- 2 * (precision * recall) / (precision + recall)
cat("Precision:", precision, "\n")

## Precision: 1

cat("Recall (Sensitivity):", recall, "\n")

## Recall (Sensitivity): 0.9895833

cat("F1-score:", f1_score, "\n")

## F1-score: 0.9947644

#ham word cloud

dtm <- DocumentTermMatrix(cleanCorpus)

# Ham word cloud
ham_indices <- which(spam_ham_df$type == "0") 
suppressWarnings(wordcloud(cleanCorpus[ham_indices], min.freq=40))

#spam word cloud

dtm <- DocumentTermMatrix(cleanCorpus)

# spam word cloud
spam_indices <- which(spam_ham_df$type == "1") 
suppressWarnings(wordcloud(cleanCorpus[spam_indices], min.freq=40))

conclusion

Test had a 98% accuracy. All of the ham intances were correctly predicted while only 1 of the spam instance was classified as not a spam

DATA 607 Project 4

angel

2024-04-13