Sentiment Analysis of Amazon Reviews

Software: R
Presentation: Shiny App
Goal: Analyzing amazon's review data to predict product sentiments

Tasks

1. Data Cleaning
       The reviews were clearned using tm package.
        Cleaning included removal of HTML tags, punctuations, extra spaces and stopwords.
2. Data Mining
       Frequency tables were built for both the normalized reviews and unnormalized reviews.
3. Data Visualization
       Interactive wordcloud was built using the rWordCloud package.
       Please run the Shiny App and have a good experience of the analysis.
4. Sentiment Analysis
       Sentiment Scores were calculated using the AFINN-111 table available online.
       Scatter charts were plotted to see the relation between user rating and calculated sentiment.

NOTE: In the data only top few reviews were considered due to computation constraints.