Prediction of spotify songs by popularity
Farah Arisya Musa, Khairi Shazwan Dollmat · AIP conference proceedings · 2024
Music popularity prediction has become one of the uprising topics in the upcoming years and it has the potential to help artists and composers into composing and releasing songs that are able to increase their popularity in the music industry. The goal of this research is to use a dataset which consists of songs from Spotify, along with the audio features to find out what are the features that can assist in predicting a track’s popularity after filtering several genres and a ran ge of popularity rates. The machine learning methods used for this research are K-Nearest Neighbours Classifier, Logistic Regression, Support Vector Machine with RBF kernel, Gradient Boosting Classifier and Random Forest Classifier that acts as a comparison method where it guides the research in evaluating which model is suitable to be used for prediction. The expected result for this research is to achieve at least more than 90% of accuracy or higher than the results achieved from other researchers. It is to be hoped that the proposed algorithms can achieve a spectacular result while providing insights on which of the following audio features may be helpful in music popularity prediction.