Watson 🔎

DISCLAIMER: This project is a Work in Progress.

Watson is a recommender system that uses Singular Value Decomposition (SVD) to find latent features in historical datasets.

Get started

python3.6 -m venv env
source env/bin/activate
pip install -r requirements.txt

The following scripts are available:

# Generate files
python generate.py
# Train and Test
python recommend.py

Goals

1. Recommendation for items for a single user based on its buying history
2. Recommendation for items, not previously bought, for a single user based on its buying historical and other simular users (Item-Item Collaborative Filtering)
3. Other users bought X also bought Y (Item-Item Collaborative Filtering)
4. Recommend similar (categorical) product (Content-Based)

Scoring

Items can be ranked based on the frequency on which items are bought. An item that was bought 10 times has a 10 times higher score compared with a product that has been bought only once.
Feature scaling is applied to normalise the scores.

Sources

1. Recommender Systems: An Introduction by Dietmar Jannach by Dietmar Jannach, Markus Zanker, Alexander Felfernig, Gerhard Friedrich.
2. Cambridge Spark - Implementing your own recommender systems in Python