First Capstone for Springboard Career Track
When consumers consider purchasing a product, they often turn to reviews and ratings submitted by other customers to determine if the purchase is worthwhile. Conversely, retailers depend on honest and accurate reviews and ratings to ensure subsequent buyers can make informed purchases. Both consumers and booksellers depend on book reviews and ratings to make informed decisions about purchases and to help with sales. Positive and negative ratings and reviews help buyers and sellers know what to spend money on and what products to avoid. Errors and inconsistencies in these assessments can directly affect sales and customer satisfaction. I propose to use features of consumer book review text to determine if reviews can predict ratings. Being able to predict ratings based on review features has multiple benefits: 1) catch errors by reviewers where they accidentally selected the wrong number of stars, 2) suggest ratings when reviewers do not provide a star rating along with their review, 3) flag confusing/incongruous review-rating pairs for revision (by reviewer) or so that they are not featured first in review lists, and potentially 4) identify and flag reviews and ratings that are ‘fake’ or jokes based on the text of the review.
| File Name | Description |
|---|---|
| README.md | Summary and Overview of Capstone |
| ProjectProposal.md | Initial Proposal of Capstone |
| DataWranglingReport.md | Documented wrangling steps |