You can check the website : http://ady24kxy10.pythonanywhere.com/
In this project, we are proposing a tool to generate ‘friendships’ based on musical preferences and playlist composition of Spotify users. The tool will take a user’s Spotify playlist(s) as input and create matches with other Spotify users based on an algorithmic compatibility of the audio features of songs within each playlist.
In this repository you will find the following directories:
-
SpotifyDataScraping: includes a python script which reads a csv file with spotify playlist links, and scrapes the data using the Spotify API ans spotypy library. It also includes the csv with the links. 1.1 Get_track_info_per_playlist.ipynb 1.2 PlaylistCollection.csv
-
FeatureEngineering: includes a python script for feature engineering and extraction as well as its inpus. Its inputs are the csv files generated from Get_track_info_per_playlist.py script. 2.1 ConsolidateVariance.ipynb 2.2 AllPlayslistData folder with csv files (inputs)
-
Models: includes 4 different python files for the 4 different models that we developed for clustering the data. Each file includes preprocessing steps. 3.1 K-means script 3.2 DBSCAN 3.3 OPTICS 3.4 Agglomerative Hierarchical Clustering
-
Pipeline: includes a python script, a csv file and a pickle model. The Script file is a machine learning pipeline which automate the end-to-end workflow of our choseen clustering model (k-means). It takes a new playlist link and user name and returns its match. The csv file has information about the clusters created with the k-means model, and the pickle file contains the model it self. 4.1 model.pkl 4.2 pipeline_data 4.3 script.ipynb