Skip to content

Machine Learning model to predict music genre based on sample of lyrics and audio (Echo-nest features)

Notifications You must be signed in to change notification settings

pippo-sci/music_gender_prediction

Repository files navigation

Music gender classification ML Model

This is the 2020-S2 COMP90049 playground for Assignment 2.

In this project the task is to perform music genre classification. Test different classification algorithms, experiment with different features, and analyse models through evaluation.

The code is in a Jupyter Notebook that it is divided in two parts that can be run independently.

Analysis

The first is the analysis part that leaded to build the final model and contains comments of the research process documented in the final paper, as other metrics.

Model development

The second has the final train, predict and evaluate function requested in the specs. They uses as arguments the dataset provided in the specs to develop the model.

Main functions:

  • train(train dataset, test dataset, train labels, test labels, random seed (default to 0)) Returns a trained model and a transformed testing dataset.

  • predict(model, test dataset transformed) Returns a list with predicted labels

  • evaluate(prediction list, test labels) Returns a classification report includin Accuracy, presicion, recall and f1- scores for each class.

Project dependencies:

  • Python 3.7
  • Pandas
  • Numpy
  • Matplotlib
  • Sklearn
  • Imblearn

Research paper

Paper link

About

Machine Learning model to predict music genre based on sample of lyrics and audio (Echo-nest features)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published