An evaluation of Non Probabilistic Classifiers and Feature Subset Selection Methods in the characterization of the popularity flag in Spotify's Track Dataset
Files: -report.pdf: Main report of problem description, methodology, findings, conclusions and references. -presentation.pdf: Presentation of the problem for 2023/2024 lectures in Machine Learning at Polytechnic University of Madrid. -Spotify_Tracks_Dataset_Processing.ipynb: Python notebook for prprocessing, data exploration, initial ML models and Wrapper approach FFS. -Data: Folder with: -dataset.csv with the original data -clean_data.csv with preprocessed data -notnorm_data.csv with preprocessed data except normalization -Weka Data folder the arff files employed for weka models are included: -knn_all.arff with all orignal variables. -knn_unifss.arff with variables after univariate FSS -knn_multfss.arff with variables after multivariate FSS