Recommender System Series

This repository contains supplementary materials for the Recommender System Series blog posts at Encora's Insights. Their goal is to give practical examples of the discussed techniques with a real-world dataset.

Installation

Use pip and requirements.txt to install the required libraries:

python -m pip install -r requirements.txt

The Notebooks were tested using Python 3.9 only, but they may work on other versions as well. The use of virtual environments is encouraged.

Dataset

We use the Amazon 2014 Product Review data. You can download it in here.

How to Start

First, follow the instructions in Preprocessing.ipynb and then proceed to explore the other Notebooks.

How this project is structured

There are two main directories in this project.

• dataset: Centralizes the storage of all the versions of the dataset, either raw or processed. It should contain the kcore_5.json and metadata.json files before running Preprocessing.ipynb.
• model: Stores the models developed in the Notebooks for easy reuse.