CollabFilter - A Python package for clustering items by similarity (Collaborative Filtering).
Vaibhavi Pai (~vaibhavi) |
Ever wondered how Amazon recommends items quite accurately while you were busy looking at another item? Or about how Netflix recommends a movie to you that you have been wanting to watch for a long time now?
This talk covers a kind of recommendation system known as Collaborative Filtering, which works by searching a large group of users and finding a smaller set with tastes similar to a particular user for whom the recommendation is made. It looks at the collection of items they have liked and combines them to create a ranked list of suggestions. Two flavours of this are item-based filtering and user-based filtering. The differences between these two can be expressed as-
1. “Users who liked this item also liked...”- Item-based Collaborative Filtering
2. “Users similar to you also liked...”- User-based Collaborative Filtering
A comprehensive open-source package in Python for Collaborative Filtering does not exist. This package, CollabFilter is our contribution to developing a recommendation system in Python. It uses a lot of state-of-art data capabilities provided by the Python Data Stack. The talk will also cover briefly on a couple of use-cases where this package was used conveniently and effectively.
Basic understanding of data science and analytics
The speaker, Vaibhavi Pai is a third year B.Tech Computer Science undergraduate student studying at The National Institute of Technology, Karnataka. She works in the field of machine learning and is keenly interested in data science and analytics. She likes working on Search Engine Optimisation and has also worked on developing games using Python in the past.