MyWebsite

Machine Learning Playlist Recommender

In this project I used a music dataset containing over 1.2 million songs, the Spotify API, Tkinter, and a Sklearn machine learning model to create a Spotify playlist based on the user’s song interests.

The dataset contains 1.2 million songs of Spotify IDs, title, artists, and a set of song traits measured numerically. The song traits analyzed are:

• Key
• Loudness
• Speechiness
• Acousticness
• Instrumentalness
• Tempo
• Dancibility

The Spotify API is used to connect the computer to the user’s Spotify account. Here is a link to the documenation: Spotify API Documentation. The built-in cosine similarity Sklearn machine learning algorithm is then used to identify similar songs that the user likes or dislikes. Cosine similarity measures the similarity of two vectors by calculating the angle between them. If the second vector is identical to the first, the second vector is given a similarity score of 1 and if they are opposite, the second vector is given a similarity score of -1.

When the program runs, a Tkinter UI appears and prompts the user to enter a title for the playlist.

A song then begins to play, the title and artisit is displayed, and the user is prompted to answer whether they like the current song.

This process continues until there are about 700 songs left, then the top 20 most similar songs to the user’s likes is created in Spotify.