/Spotify-Recommendation-System

Recommender system using XGBOOST, Neural_Network, Ensemble and LGBM

Primary LanguageJupyter Notebook

Spotify Recommendation System

Introduction

This repository contains a Spotify recommendation system that uses various machine learning models to predict whether a user will like a song or not. The dataset used in this analysis was collected from the Spotify API, which provides access to a vast repository of music metadata.

Models

XGBoost

  • AUC-ROC: 0.974
  • Likelihood of liking: 0.082

Neural Network

  • Likelihood of liking by user: 0.290

Ensemble Models

Random Forest

  • Accuracy Score: 0.9230769230769231
  • ROC AUC Score: 0.9236842105263158

Extra Tree Classifier

  • Accuracy Score: 0.9487179487179487
  • ROC AUC Score: 0.9486842105263158

Bagging Classifier

  • Accuracy Score: 0.8717948717948718
  • ROC AUC Score: 0.8723684210526317

AdaBoost Classifier

  • Accuracy Score: 0.8717948717948718
  • ROC AUC Score: 0.8723684210526317

LightGBM Classifier

  • Accuracy Score: 0.8717948717948718
  • ROC AUC Score: 0.8723684210526317

Conclusion

This repository demonstrates the effectiveness of various machine learning models in predicting whether a user will like a song or not. The results show that the XGBoost model performed well in terms of AUC-ROC and likelihood of liking. The ensemble models also showed promising results, with the Random Forest model performing well in terms of accuracy and ROC AUC score.