The machine learning used to power this app is all located in the blockbox-api repo which can be found at https://github.com/jayshrivastava/blockbox-api.
https://blockbox-web.herokuapp.com/
This was created to demonstrate how retail insights, such as product recommendations, can be generated through simple machine learning algorithms. All ML algorithms are written in pure TypeScript and find recommendations based on anonymous ratings data. The app allows users to rate movies and generate recommendations for themselves.
Recommendations are generated using collaborative filtering. First, users are compared by cosine similarity (comparing the vectors of movie ratings by angle). Then, users are recommended items based on what similar users rated or 'watched'. Each relevant movie that may be of interest to user A is given a score based on how similar user A is to other users and how those users rated said movies. From there, the top scoring movies are recommended to user A.
For more info, see this guide. The models used to build this app follow the same idea, but with a few tweaks.
The initial ratings/movies data is from the MovieLens Latest Dataset with movies dated from 1990 to 09/2018. The exact seed data can be found in the blockbox-infra repository which also contains the script to seed the blockbox Mongo DB.
As users rate movies, their data is also streamed in real time to MongoDB, so the dataset is consistently updated.
Also here are some good collaborative filtering datasets.
Node v. 12.4.0
Yarn v. 1.16.0
This project was bootstrapped with Create React App.
You will need to add certain environment variables to hook this up to the blockbox-api.
yarn install
Installs dependencies
yarn run build
Builds the app for production to the build folder.
It correctly bundles React in production mode and optimizes the build for the best performance.
The build is minified and the filenames include the hashes.
Your app is ready to be deployed!
See the section about deployment for more information.
yarn start
Runs the app in the development mode.
Open http://localhost:3000 to view it in the browser.
The page will reload if you make edits.
You will also see any lint errors in the console.