Project in data analysis. Imagine you were working in Yahoo Music or Youtube Music and your task was to find a very simple and quick way of separating all the artists into N groups (genre-wise, for example). You are also given a graph with artists as nodes. The more often two artists are listened by the same user, the bigger is the weight value of the vertice.
Spectral Clustering Algorithm (according to Ng, Jordan and Weiss, https://papers.nips.cc/paper/2092-on-spectral-clustering-analysis-and-an-algorithm.pdf) was implemented on the Yahoo music dataset and compared with other clustering algorithms (Hierarchical Clustering and DBSCAN). Jupyter Notebook was used to write the code and visualize the results.
To reproduce the results, just clone the repo, unzip all the archieved files and launch the notebook