This project is done in accordance to the Mini-Project for the Machine Learning Course (COMP 484) of Kathmandu University, Department of Computer Science and Engineering.
Contributors:
- Romit Khanal (Me)
- Sandesh Balaya Shrestha
- Dr. Bal Krishna Bal (Instructor)
The idea behind MusicLearn(ML) is to build a model to classify songs based on their lyrics by using naive Bayes classifier. We have used 10-fold cross validation to train in the dataset. The dataset has been downloaded from the Million Song Dataset Subset. We then filtered this dataset to a 1000-song training dataset and 200-song test dataset.
Prerequisites:
- Anaconda or Jupyter Notebook
To run this project, you would need to follow the given steps:
- Download the project either through a zip file or using the command:
git clone
from the command line. - Open the downloaded folder and navigate to the file from your Jupyter Notebook
MusicLearn.ipynb
- Run the full notebook.
- Verify the graphs in the folder named
images
.