Make sure to unzip the data folder before running the python notebook.

This is a very simple song recommendation system.

We have used the FMA- Music analysis dataset for this task. We used the acoustic features of the song for classification.

Since we wanted to make it a binary classification for a starting project, we have created the labels by using 1 and 0. 1 showing that user liked the song and 0 showing that the user disliked the song.

We have created the labels based on a user interest of genres, that is randomized on every run. We have also added a bias for the number of 1's that are assigned to the songs in a specific genre. This bias was added because we understand that one cannot like all the songs in a specific genre.

We have used three different models for classification namely k-Nearest Neighbors, Linear Regression and Multi Layer Perceptron (MLP).

We have computer accuracy and confusion matrix for all the models and also plotted the graphs of the types of songs that are suggested to the user.