Problem Set 1 Problem 1

[TSoellinger]

I.Introduction

   File Name: Email Classifier.R
   Due: 02/28/14 at 5:00 p.m.
   Author: Tyler Soellinger
   RE: Replicate the Spam versus Ham Classifier Example from Chapter 3 of Machine Learning for Hackers
   Data Used: Email messages contained in data/ directory, source: http://spamassassin.apache.org/publiccorpus

II.Code

1. Load Libraries

library(ggplot2)   
library(tm)

2.Set Global Paths For Email Archives

spam.path <- file.path("data", "spam")   
spam2.path <- file.path ("data", "spam2")   
spam2.path <- file.path ("data", "spam_2")   
easyham.path <- file.path ("data", "easy_ham")   
easyham2.path <- file.path ("data", "easy_ham_2")   
hardham.path <- file.path ("data", "hard_ham")  
hardham2.path <- file.path ("data", "hard_ham_2")  

3. Create Motivating Plot

x <- runif(1000, 0, 40)   
y1 <- cbind(runif(100, 0, 10), 1)   
y2 <- cbind(runif(800, 10, 30), 2)   
y3 <- cbind(runif(100, 30, 40), 1)   

val <- data.frame(cbind(x, rbind(y1, y2, y3)), stringsAsFactors = TRUE)   

ex1 <- ggplot(val, aes(x, V2)) +     
  geom_jitter(aes(shape = as.factor (V3)), position = position_jitter(height = 2)) +    
  scale_shape_discrete(guide = "none", solid = FALSE) +    
  geom_hline(aes(yintercept = c(10,30))) +    
  theme_bw() + xlab("X") + ylab("Y")      

ggsave(plot = ex1, filename = file.path("images", "00_Ex1.pdf"), height = 10, width = 10)   

4. Return A Single Element Vector Of Just The Email Body With Words As Features

get.msg <- function(path) 
  {
    con <- file(path, open = "rb", encoding = "latin1")
    text <- readLines (con)

    The Mesage Always Begins After The First Full Line Break

    msg <- text[seq(which(text == "") [1] + 1, length(text), 1)]
    close(con)
    return(paste(msg, collapse = "\n"))
  }

5. Create A TermDocumentMatrix (TDM) From The Corpous Of SPAM Email

    This TDM Creates The Feature Set Used To Train Our Classifier

get.tdm <- function(doc.vec) 
  {
    control <- list(stopwords = TRUE, removePunctuation = TRUE, removeNumbers = TRUE, minDocFreq = 2)
    doc.corpus <- Corpus(VectorSource(doc.vec))
    doc.dtm <- TermDocumentMatrix(doc.corpus, control)
    return(doc.dtm)
  }

    This Function Takes A File Path To An Email File And A String, The Term Parameter, And Returns The Count Of That Term In The Email Body

count.word <- function(path, term) 
  {
    msg <- get.msg(path)
    msg.corpus <- Corpus(VectorSource(msg))
    control <- list(stopwords = TRUE, removePunctuation = TRUE, removeNumbers = TRUE)
    msg.tdm <- TermDocumentMatrix(msg.corpus, control)
    word.freq <- rowSums(as.matrix(msg.tdm))
    term.freq <- word.freq[which(names(word.freq) == term)]

   When Nothing Is Found, "Ifelse" Is Used To Produce "0" Instead Of "NA"

    return(ifelse(length(term.freq) > 0, term.freq, 0))
  }

6. This Does The Heavy Lifting For Classifying Email By Taking Two Required Parameters; A File Path To An Email To Classify, And A Data Frame Of The Trained Data. This Also Takes Two Optional Parameters: First, A Prior Over The Probability That An Email Is SPAM, Which We Set To 0.5, And Constant Value For The Probability On Words In The Email That Are Not In Our Training Data

   This Function Returns The Naive Bayes Probability That The Given Email Is SPAM

classify.email <- function(path, training.df, prior = 0.5, c = 1e-6) 
  {

   Get The Email Text In A Workable Format

    msg <- get.msg(path)  
    msg.tdm <- get.tdm(msg)  
    msg.freq <- rowSums(as.matrix(msg.tdm))  

   Find Intersection Of Words

    msg.match <- intersect(names(msg.freq), training.df$term)  

   Perform The Naive Bayes Calculation

    if(length(msg.match) < 1)
  {
    return(prior * c ^ (length(msg.freq)))
  }
    else
  {
    match.probs <- training.df$occurrence[match(msg.match, training.df$term)]
    return(prior * prod(match.probs) * c ^ (length(msg.freq) - length(msg.match)))
  }
  }

7. Perform The Classifications

   (1) Create A Document Corpus For Spam Messages

      Place All The SPAM-y Email Into A Single Vector

spam.docs <- dir(spam.path)  
spam.docs <- spam.docs[which(spam.docs != "cmds")]  
all.spam <- sapply(spam.docs, function(p) get.msg(file.path(spam.path, p)))  

      Create A DTM For That Vector

spam.tdm <- get.tdm(all.spam)

      Create A Data Frame With The Feature Set From The Training SPAM Data

spam.matrix <- as.matrix(spam.tdm)  
spam.counts <- rowSums(spam.matrix)  
spam.df <- data.frame(cbind(names(spam.counts), as.numeric(spam.counts)), stringsAsFactors = FALSE)  
names(spam.df) <- c("term", "frequency")  
spam.df$frequency <- as.numeric(spam.df$frequency)  

spam.occurrence <- sapply(1:nrow(spam.matrix), function (i)  
  {  
    length(which(spam.matrix[i, ] > 0)) / ncol(spam.matrix)  
  })  

spam.density <- spam.df$frequency / sum(spam.df$frequency)  

      Add The Term Density And Occurrence Rate

spam.df <- transform(spam.df, density = spam.density, occurrence = spam.occurrence)  

   (2) Repeat For EasyHam Email

easyham.docs <- dir(easyham.path)
easyham.docs <- easyham.docs[which(easyham.docs != "cmds")]
all.easyham <- sapply(easyham.docs[1:length(spam.docs)], function(p) get.msg(file.path(easyham.path, p)))

easyham.tdm <- get.tdm(all.easyham)

easyham.matrix <- as.matrix(easyham.tdm)
easyham.counts <- rowSums(easyham.matrix)
easyham.df <- data.frame(cbind(names(easyham.counts), as.numeric(easyham.counts)), stringsAsFactors = FALSE)
names(easyham.df) <- c("term", "frequency")
easyham.df$frequency <- as.numeric(easyham.df$frequency)

easyham.occurrence <- sapply(1:nrow(easyham.matrix), function(i)
  {
    length(which(easyham.matrix[i, ] > 0)) / ncol(easyham.matrix)
  })

easyham.density <- easyham.df$frequency / sum(easyham.df$frequency)
easyham.df <- transform(easyham.df, density = easyham.density, occurrence = easyham.occurrence)

8. Run Classifier Against Hard Ham

hardham.docs <- dir(hardham.path)
hardham.docs <- hardham.docs[which(hardham.docs != "cmds")]

hardham.spamtest <- sapply(hardham.docs, function (p) classify.email(file.path(hardham.path, p), training.df = spam.df))  

hardham.hamtest <- sapply(hardham.docs, function(p) classify.email(file.path(hardham.path, p), training.df = easyham.df))

hardham.res <- ifelse(hardham.spamtest > hardham.hamtest, TRUE, FALSE)
summary(hardham.res)

Mode	FALSE	TRUE	NA's
Logical	243	6	0

9. Find Counts Of Just Terms 'html' And 'table' In All SPAM And EASYHAM Docs, And Create Figure

html.spam <- sapply(spam.docs, function(p) count.word(file.path(spam.path, p), "html"))  
table.spam <- sapply(spam.docs, function(p) count.word(file.path(spam.path, p), "table"))  
spam.init <- cbind(html.spam, table.spam, "SPAM")  

html.easyham <- sapply(easyham.docs, function(p) count.word(file.path(easyham.path, p), "html"))
table.easyham <- sapply(easyham.docs, function(p) count.word(file.path(easyham.path, p), "table"))
easyham.init <- cbind(html.easyham, table.easyham, "EASYHAM")

init.df <- data.frame(rbind(spam.init, easyham.init), stringsAsFactors = FALSE)
names(init.df) <- c("html", "table", "type")
init.df$html <- as.numeric(init.df$html)
init.df$table <- as.numeric(init.df$table)
init.df$type <- as.factor(init.df$type)

    Plot 1

init.plot1 <- ggplot(init.df, aes(x = html, y = table)) + 
  geom_point(aes(shape = type)) + 
  scale_shape_manual(values = c("SPAM" = 1, "EASYHAM" = 3), name = "Email Type") + 
  xlab("Frequency of 'html'") + 
  ylab("Frequency of 'table'") + 
  stat_abline(yintersept = 0, slope = 1) + theme_bw()  

ggsave(plot = init.plot1, filename = file.path("images", "01_init_plot1.pdf"), width = 10, height = 10)  

    Plot 2

init.plot2 <- ggplot(init.df, aes(x = html, y = table)) + 
  geom_point(aes(shape = type), position = "jitter") + 
  scale_shape_manual(values = c("SPAM" = 1, "EASYHAM" = 3), name = "Email Type") + 
  xlab("Frequency of 'html'") + 
  ylab("Frequency of 'table'") + 
  stat_abline(yintersept = 0, slope = 1) + theme_bw()  

ggsave(plot = init.plot2, filename = file.path("images", "02_init_plot2.pdf"), width = 10, height = 10)  

10. Classify HARDHAM Data Using The Classifier Developed Above

spam.classifier <- function(path)  
  {  
    pr.spam <- classify.email(path, spam.df)  
    pr.ham <- classify.email(path, easyham.df)  
    return(c(pr.spam, pr.ham, ifelse(pr.spam > pr.ham, 1, 0)))  
  }  

11. Get Lists Of All The Email Messages

easyham2.docs <- dir(easyham2.path)  
easyham2.docs <- easyham2.docs[which(easyham2.docs !="cmds")]  

hardham2.docs <- dir(hardham2.path)  
hardham2.docs <- hardham2.docs[which(hardham2.docs != "cmds")]  

spam2.docs <- dir(spam2.path)  
spam2.docs <- spam2.docs[which(spam2.docs != "cmds")]  

Classify All Of Them

easyham2.class <-suppressWarnings(lapply(easyham2.docs, function(p) 
  { 
    spam.classifier(file.path(easyham2.path, p)) 
  }))  

hardham2.class <- suppressWarnings(lapply(hardham2.docs, function(p) 
  { 
    spam.classifier(file.path(hardham2.path, p)) 
  }))  

spam2.class <- suppressWarnings(lapply(spam2.docs, function(p) 
  { 
    spam.classifier(file.path(spam2.path, p)) 
  }))  

12. Create A Single, Final Data Frame With All Classifications Of Data

easyham2.matrix <- do.call(rbind, easyham2.class)
easyham2.final <- cbind(easyham2.matrix, "EASYHAM")

hardham2.matrix <- do.call(rbind, hardham2.class)
hardham2.final <- cbind(hardham2.matrix, "HARDHAM")

spam2.matrix <- do.call(rbind, spam2.class)
spam2.final <- cbind(spam2.matrix, "SPAM")

class.matrix <- rbind(easyham2.final, hardham2.final, spam2.final)
class.df <- data.frame(class.matrix, stringsAsFactors = FALSE)
names(class.df) <- c("Pr.SPAM" ,"Pr.HAM", "Class", "Type")

class.df$Pr.SPAM <- as.numeric(class.df$Pr.SPAM)
class.df$Pr.HAM <- as.numeric(class.df$Pr.HAM)
class.df$Class <- as.logical(as.numeric(class.df$Class))
class.df$Type <- as.factor(class.df$Type)

13. Create Final Plot Of Results

class.plot <- ggplot(class.df, aes(x = log(Pr.HAM), log(Pr.SPAM))) + 
  geom_point(aes(shape = Type, alpha = 0.5)) + 
  stat_abline(yintercept = 0, slope = 1) + 
  scale_shape_manual(values = c("EASYHAM" = 1, "HARDHAM" = 2, "SPAM" = 3), name = "Email Type") + 
  scale_alpha(guide = "none") + 
  xlab("log[Pr(HAM)]") + 
  ylab("log[Pr(SPAM)]") + 
  theme_bw() + 
  theme(axis.text.x = element_blank(), axis.text.y = element_blank())

ggsave(plot = class.plot, filename = file.path("images", "03_final_classification.pdf"), height = 10, width = 10)

get.results <- function(bool.vector)
  {
    results <- c(length(bool.vector[which(bool.vector == FALSE)]) / length(bool.vector), length(bool.vector[which(bool.vector == TRUE)]) / length(bool.vector))
    return(results)
  }

14. Save Results As A Table

easyham2.col <- get.results(subset(class.df, Type == "EASYHAM")$Class)
hardham2.col <- get.results(subset(class.df, Type == "HARDHAM")$Class)
spam2.col <- get.results(subset(class.df, Type == "SPAM")$Class)

class.res <-rbind(easyham2.col, hardham2.col, spam2.col)
colnames(class.res) <- c("NOT SPAM", "SPAM")

15. Result Outputs

print(class.res)

Class	NOT SPAM	SPAM
easyham2.col	0.9871429	0.01285714
hardham2.col	0.9677419	0.03225806
spam2.col	0.4631353	0.53686471

16. Save The Training Data For Potential Future Use

write.csv(spam.df, "data/spam_df.csv", row.names = FALSE)
write.csv(easyham.df, "data/easyham_df.csv", row.names = FALSE)

END