Recommendation System - Hepsiburada
Table of Contents
About
In this project, I built a recommendation system for an online grocery store. The idea is building a related-products recommendation system for the cart page of an online store. Meaning that when a customer is landed on the cart page, a list of recommended products should be provided.
Datasets
- Events.json: contains events of type 'cart', adding a product to the cart.
- Meta.json: contains info about the products.
Dependencies
To run the notebook:
- pandas==1.1.3
- gensim==4.0.1
- sklearn
- json
- string
To run the API example:
- pandas==1.1.3
- requests==2.21.0
- Flask==1.1.2
- json
Usage
The repo contains a comprehensive notebook that I recommend starting with.
There is also a Design Document provided, which discusses some of the things in the notebook in more detials and also the futuer work.
Other than that, an API is added for the sake of providing an example as it was required. Check it out if you are interested, but the notebook is already detailed enough to help in understanding the project.
Running The API Example
To run the API Example:
- first need to run the notebook. Running the notebook will invove pickling some objects that are used on the server. Make sure the datasets (json files) exist in the cwd in a directory called data.
- start the flask server and leave it open.
python3 api_example/flask_api.py
- run the examples.py script and wait for a response from the server.
python3 api_example/examples.py