Predicting Amazon Review Helpfulness
View on GitHub Applied Data Analytics class, 2018.
Our group wanted to find out if we could predict the “helpfulness” of a product review on Amazon. We explored the relationships between helpfulness and length, star-rating, words used, and trends over time. We found that the review length, star-rating, and the words used in a review were the most telling factors. Furthermore, a new classifier is not necessarily needed for each product category. We found that the same words that indicate that a beauty product review is helpful can also tell us if an automotive product or baby product review is helpful.
You can read our final report here