A Recommendation engine for an e-commerce use case that provides recommendations to users based on their purchase history. It has been built using the matrix factorization algorithm.
Installation
The dependencies required to run the code at your end can be installed from the requirements.txt file within your virtual environment using
pip install -r requirements.txt
The app has been developed using H2o.ai Wave. Please refer to their detailed documentation to setup and run the Wave server on your machine. Once that's done, you can easily run any wave app .
Development
This notebook converts the raw data stored in json.gz of different categories into a single csv file. It also combines the metadata of different categories which is required by our wave application for displaying recommendations to the user.
This is where the model is trained,hyperparameters are tuned and the final model is saved to be used by the wave app to provide recommendations. The model has been trained using the Matrix Factorization module of the h2o4gpu package.
The web application uses the model built in the previous notebook and other output files to provide product recommendations relevant for each selected user.
Running your Wave app
After setting up your environment with the dependencies installed, navigate to the rec-system directory and execute the following command
wave run app.py
Now use any web browser of your choice to view the app at : localhost:10101/rec
Do check out the following blog post for a more detailed overview:
https://www.h2o.ai/blog/what-are-we-buying-today/
Citations
Justifying recommendations using distantly-labeled reviews and fined-grained aspects Jianmo Ni, Jiacheng Li, Julian McAuley Empirical Methods in Natural Language Processing (EMNLP), 2019 pdf