/artist_recommender

Recommends the KNearestNeighbors for a given artist/musician. Try it out: http://artist-recommender.herokuapp.com/

Primary LanguageJupyter Notebook

Purpose 🎯

This is a very small project experimenting with the KNeighborsRegressor from the SkitLearn library. The purpose is to showcase a clean workflow analyzing a data set of ~30,000 artists provided by this kaggle source.

Contents 📒

This repo mainly focusses on a notebook analyzing the given data and building a small model to make find similar musicians for a given input musician. The notebook can be found here.

It is structured in the following way:

  1. Introduction
  2. Imports
  3. The Dataset
  4. Provided Features
  5. Preprocessing
  6. Model
  7. Recommender

The Recommender 🥁

The final recommender is also pushed to Heroku and available for everyone to try. Please follow this link to get there: http://artist-recommender.herokuapp.com/. The recommender is case, special-char and interpunction insensitive and therefore quite stable for a small project like this. Although, it will only recognize fully written words and not auto-fill them (like 'Emine' or 'beatle'). Try it out yourself and have fun! 🥁

The app is run on a Streamlit Framework and hosted by heroku. Longer loading times might occur.