rm(list=ls())

data<- read.csv("ClothReview.csv")

library(dplyr)

data=sample_n(data,100)

# install.packages("readr")

library(readr)

#Text mining packages

# install.packages("tm")

library(tm)

# install.packages("SnowballC")

library(SnowballC)

corpus = VCorpus(VectorSource(data$Review.Text))

corpus[[1]][1]

data$Recommended.IND[1]

corpus = tm_map(corpus,PlainTextDocument)

corpus = tm_map(corpus,content_transformer(tolower))

corpus[[1]][1]

corpus = tm_map(corpus,removePunctuation)

corpus[[1]][1]
corpus = tm_map(corpus, removeWords, c("cloth", stopwords("english")))

corpus = tm_map(corpus, stemDocument)

frequencies = DocumentTermMatrix(corpus)

sparse = removeSparseTerms(frequencies, 0.99)

tSparse = as.data.frame(as.matrix(sparse))

colnames(tSparse) = make.names(colnames(tSparse))

tSparse$recommended = data$Recommended.IND

# install.packages("caTools")

library(caTools)

split = sample.split(tSparse$recommended, SplitRatio = 0.8)

trainSparse = subset(tSparse, split==TRUE)

testSparse = subset(tSparse, split==FALSE)

library(randomForest)

trainSparse$recommended = as.factor(trainSparse$recommended)

testSparse$recommended= as.factor(testSparse$recommended)

RF_model = randomForest(recommended ~ ., data=trainSparse)

predictRF = predict(RF_model, newdata=testSparse)

library(caret)
confusionMatrix(table(predictRF,testSparse$recommended),positive='1')