Music Genre Classification

Introduction

We aim in this project to classify music based on tabular dataset
The dataset contains data about 15000 song , each song has 17 attribute to describe it , such as loudness , acousticness , instrumentalness and more ..

Discovering the data

From the next histogram grid , we can get many insights:
1-The Id column has no useful impact,2-There are some skewed feature ,3-There are some feature with discrete values
From the next heatmap , we can notice some linear corrleation between some feature like loudness vs. energy
We can see from the next barplot that we have a lot of outliers that we should handle them by flooring and capping , and some transformation methods

Pre-Processing Steps

First,we dropped the duplicates based on the track name & the artist name , then we handled the outlieres , then we applied some reformatting on some features , then we applied some transformation to handle the skewed data

Evaluation & Model Selection

We applied different machine learning models ,the cat-boost got the best result about 0.70 F1 Score