Music

Machine Learning - Music Recommender System Using Spotify Dataset


Tags: #Recommender #Spotify #EDA #MachineLearning #Music

Background

Spotify is one of the music streaming provider with over 422 million active users and 182 million paying subscribers as of June 2022. By the large number of their user base, Spotify generates a huge database from their users and the amount of their database is growing exponentially each day. The popular music streaming also provides various recommendations to the users based on their data or other users with the same preferences such as artist, album, genre, playlist and many more to provide an accurate personal user experience.

Solution

Some companies utilize a recommender system as a decision making strategy for them under complex information environments, which was defined for assisting the company to make choices when there is insufficient or over sufficient data provided. In order to manage such complex data environment, company tend to use collaborative filtering to build the recommendation model.

Exploratory Data Analysis

Analytico Asia utilized the cosine similarities to filter and build the model by identifying other users in the Spotify database with similar taste and leverage their opinion to recommend songs to the active user.