Recommending True Love with Non-Negative Matrix Factorization
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data
.gitignore
README.md
article6_eda.py
article6_train_recommender.py
article6_use_recommender.py

README.md

Recommending True Love with Non-Negative Matrix Factorization

Like PCA, Non-Negative Matrix Factorization (NNMF) reduces the dimensionality of data. Unlike PCA, the component features it creates are more easily interpretable. For the purpose of our "love recommender" we are going to use NNMF to kill two birds with one stone: first we will reduce the massive dimensionality of our NLP, and second we will use the similarities among these reduced features to find good matches, that can be interpreted and explained.

Read the full article here: https://datajenius.com/articles/recommending-true-love-with-non-negative-matrix-factorization

Technical Note

You must add profiles.csv to the /data folder before all of these scripts will work.

This file is available here: https://github.com/rudeboybert/JSE_OkCupid/blob/master/profiles.csv.zip