The notebook offers a way to solve a problem stated in Fang Zhou, Claire Q and Ross. D. King, Predicting the Geographical Origin of Music, ICDM, 2014. The solution is based on classification using random forest classifier. To measure the accuracy of predictions, the custom geographical accuracy metric was designed. The metric is described in the notebook and its coefficients can be found in the attached .txt file.
The method uses scikit-learn library's implementations of machine learning methods and pandas' tools for handling data.
The data set was created by Fang Zhou, The University of Nottinghan, Ningbo, China and was shared here.