https://github.com/lene/tavsiye

A very basic recommendation engine implemented in Python.
https://github.com/lene/tavsiye

Last synced: 5 months ago
JSON representation

A very basic recommendation engine implemented in Python.

• Host: GitHub
• URL: https://github.com/lene/tavsiye
• Owner: lene
• License: gpl-3.0
• Created: 2015-04-08T14:59:46.000Z (about 11 years ago)
• Default Branch: master
• Last Pushed: 2018-03-05T12:20:07.000Z (over 8 years ago)
• Last Synced: 2025-03-21T22:02:58.163Z (over 1 year ago)
• Language: Python
• Homepage:
• Size: 43.9 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
A very basic recommendation engine implemented in Python.

Usage:

    `python recommender.py [-h|-f FILENAME|--user USER|--set-comparison SET_COMPARISON|--cutoff CUTOFF]`

Running test suite:

    `python test_recommender.py && mypy . && flake8 --max-line-length=100`

Implementation thoughts:

- The entire recommendation algorithm can be found in compare_sets.py.

- I found it useful to plug different algorithms for computing the similarity of sets into the

 recommendation engine. The other comparison algorithms can be found in alternative_methods.py.

- In principle, nothing stops you from plugging custom objects as a similarity_matrix into

 the similar_users() and recommendations() functions. (I don't supply an example for this though.)

 You might want to do that because the similarity_matrix I calculate grows is O(N^2) in the number

 of users - in real world settings you'd want to optimize this.

- You have to decide on a numeric value for the similarity of two sets of liked products to

 consider two users similar in taste. I plugged that value as parameter 'cutoff' into the relevant

 functions.

- I am not sure the MinHash-algorithm I used is stable - I sort of whipped it up. It is hard to

 test stochastic algorithms, I considered this outside the scope of this exercise.