Scribd
Upload
154 views3 pages

Naive Bayes Text Classification Guide

This document summarizes the process of building a Naive Bayes classifier to perform text classification on the 20 Newsgroups dataset. It involves preprocessing the text by converting to lowercase, removing punctuation, numbers and stopwords. Then it creates document-term matrices for training and test data, builds a Naive Bayes model on the training data, makes predictions on the test data and evaluates the model using a confusion matrix.

Uploaded by

brahmesh_sm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
154 views3 pages

Naive Bayes Text Classification Guide

This document summarizes the process of building a Naive Bayes classifier to perform text classification on the 20 Newsgroups dataset. It involves preprocessing the text by converting to lowercase, removing punctuation, numbers and stopwords. Then it creates document-term matrices for training and test data, builds a Naive Bayes model on the training data, makes predictions on the test data and evaluates the model using a confusion matrix.

Uploaded by

brahmesh_sm
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

# Text

classification
using a Naive
Bayes scheme
# Data : 20 Newsgroups
# Download link : [Link]

# Load all the required libraries. Note : Packages need to be installed first.
library(dplyr)
library(caret)
library(tm)
library(RTextTools)
library(doMC)
library(e1071)
registerDoMC(cores=detectCores())
# Load data.
# We will use the 'train-all-terms' file which contains over 11300 messages.
# Read file as a dataframe
[Link] <- [Link]("[Link]", header=FALSE, sep="\t", quote="",
stringsAsFactors=FALSE, [Link] = c("topic", "text"))

# Preview the dataframe


# head([Link]) # or use View([Link])
# How many messages do each of the 20 categories contain?
table([Link]$topic)
# Read topic variable as a factor variable
[Link]$topic <- [Link]([Link]$topic)

# Randomize : Shuffle rows randomly.


[Link](2016)
[Link] <- [Link][sample(nrow([Link])), ]
[Link] <- [Link][sample(nrow([Link])), ]
# Create corpus of the entire text
corpus <- Corpus(VectorSource([Link]$text))

# Total size of the corpus


length(corpus)

# Inspect the corpus


inspect(corpus[1:5])
# Tidy up the corpus using 'tm_map' function. Make the following transformations on
the corpus : change to lower case, removing numbers,
# punctuation and white space. We also eliminate common english stop words like
"his", "our", "hadn't", couldn't", etc using the
# stopwords() function.
# Use 'dplyr' package's excellent pipe utility to do this neatly
[Link] <- corpus %>%
tm_map(content_transformer(tolower)) %>%
tm_map(removePunctuation) %>%
tm_map(removeNumbers) %>%
tm_map(removeWords, stopwords(kind="en")) %>%
tm_map(stripWhitespace)
# Create document term matrix
dtm <- DocumentTermMatrix([Link])
dim(dtm)
# Create a 75:25 data partition. Note : 5000 (~50% of the entire set) messages were
used for this analysis.

[Link] <- [Link][1:8470,]


[Link] <- [Link][8471:11293,]

[Link] <- dtm[1:8470,]


[Link] <- dtm[8471:11293,]
dim([Link])
[Link] <- [Link][1:8470]
[Link] <- [Link][8471:11293]
# Find frequent words which appear five times or more

fivefreq <- findFreqTerms([Link], 5)


length(fivefreq)
dim([Link])
# Build dtm using fivefreq words only. Reduce number of features to
length(fivefreq)
[Link]( [Link] <- DocumentTermMatrix([Link], control =
list(dictionary=fivefreq)) )
[Link]( [Link] <- DocumentTermMatrix([Link], control =
list(dictionary=fivefreq)) )
# converting word counts (0 or more) to presence or absense (yes or no) for each
word
convert_count <- function(x) {
y <- ifelse(x > 0, 1,0)
y <- factor(y, levels=c(0,1), labels=c("No", "Yes"))
y
}
# Apply yes/no function to get final training and testing dtms
[Link]( [Link] <- apply([Link], 2, convert_count) )
[Link] ( [Link] <- apply([Link], 2, convert_count) )
# Build the NB classifier
[Link] (ngclassifier <- naiveBayes([Link], [Link]$topic))

# Make predictions on the test set


[Link]( predictions <- predict(ngclassifier, newdata=[Link]) )
predictions
cm <- confusionMatrix(predictions, [Link]$topic )
cm

You might also like