Building a sentiment analysis model to categorize words based on their sentiment. That is, whether the words are positive or negative and further analysis.
Dataset/Package: janeaustenr
First, I made use of the tidytext package that consists of sentiment lexicons present in the dataset of sentiments.
Among the various lexicons, I have retrieved the bing lexicon, which classifies the sentiments into a binary category of negative or positive.
I imported the libraries janeaustenr, stringr and tidytext. The janeaustenr package which provided me with full texts for Jane Austen's 6 completed novels, ready for text analysis. These novels are "Sense and Sensibility", "Pride and Prejudice", "Mansfield Park", "Emma", "Northanger Abbey", and "Persuasion".tidytext allowed me to perform efficient text analysis on the data.
After performing the tidy operation on the text such that each row contains a single word, I used the book 'Emma' and derived its words to implement the sentiment analysis model.
I then segregated the data into separate columns of positive and negative sentiments. After that, using mutate(), I calculated the total sentiment, that is, the difference between positive and negative sentiment.
First visualisation: the words present in the book “Emma” based on their corresponding positive and negative scores. I counted the most common positive and negative words that are present in the novel.
Second visualisation: sentiment score. I plotted the scores along the axis that is labeled with both positive as well as negative words.
Final visualisation: I created a wordcloud that described the most recurring positive and negative words.
Cr:DataFlair