team_under_cover
April 5, 2018 OVERVIEW
- Project Background and Description Can you really judge a book by its cover? We intend to use machine learning to find out! We will use machine learning to create models for the best and worst amazon reviewed book covers and create a model that will allow a user to upload a book cover image and receive a predicted review score.
- Project Scope We will need a script to query our amazon data and strip the url from the review metadata. Then, we will need to use Google’s Vision API to run analysis on the images. Once we have the data of various aspects of each image, we will train our models with what characteristics are indicative of a high and low score.
- High-Level Requirements Python, HTML, CSS, Java, Google Vision API, Amazon dataset
- Data Sources
- Amazon Dataset
- Google Cloud Vision API
- Deliverables HTML Webpage that accepts a user-uploaded file and predicts either rating or genre based on cover image.
- Sample Charts
- Sketch Title Page with project description and links to 3 pages of charts
Sentiment Analysis Page with sentiments over time Year vs Compound score, grouped by genre Word Analysis page with word count and unique word count data Word to unique word ratio, either hear map or layered historgram By Genre and Song page that allows drill down Attempting checkboxes to filter chart