Skip to content

Latest commit

 

History

History
67 lines (45 loc) · 2.16 KB

README.md

File metadata and controls

67 lines (45 loc) · 2.16 KB
Raw

MR. HyPE: Music Recommendation
based on Hyper-dimensional Python Embedding

ECE143 Group 20

Installation

Requires python 3.3+

Some main third-party modules:

  • pandas 1.1.4
  • sklearn 0.0
  • plotly 4.13.0
  • spotipy 2.16.1

Clone the repository using

git clone https://github.com/ArthDh/ECE-143/

Create a python virtual environment

python -m venv env

Activate the environment

source  env/bin/activate

Install dependencies

pip install requirements.txt

Deactivate when done making changes

deactivate

Usage

All the data we used and generated are stored in data folder, and the clustering model we generated are stored in model folder.

Data Visualization

Visualizations of the correlation between raw features from dataset are presented in final notebook in notebooks folder.
For a better viewing experience and interactibility view notebook here nbviewer

Clustering

We used K-mean for clustering and all visualizations regarding K-mean are also showed in final notebook. The original data used for hyper dimensional visualization are stoed in data folder, named df_cleaned.tsv and df_cleaned_genre_10.tsv.

Hyper Dimensional Visualization

PCA

PCA

UMAP

UMAP

In order to see the 3D visualization of the dataset with predicted genres and artist names as index, use the following link:
Embedding Visualizer

Artist Recommendation

Go to the final notebook, change the sample name to your favorite artist and change the lim to the length of the recommendation list you want, and run the notebook.