Song popularity prediction using machine learning

• Main Author: Feng, Zhilei
• Other Authors: Wang Lipo
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Machine learning
• Online Access: https://hdl.handle.net/10356/173954
• Institution: Nanyang Technological University

Forecasting the future popularity of songs holds significant appeal for the music industry. Potential applications encompass evaluating the prospects of a novel song, developing automated songwriting aides, and designing song recommendation systems. There are many factors that influence a song’s popularity, and the problem of predicting the future of a song to be released is even more difficult to solve. The rapid development of machine learning models provides a feasible solution to this problem. This dissertation uses multiple machine learning models to predict song popularity. Then the dataset is created for research from Spotify. The machine learning model is tested using two methods: machine learning algorithm and deep learning model. Among the seven machine learning algorithms used in the research, XGBOOST achieved the best results. After using the deep learning model, the model trained through the convolutional neural network achieved higher results than XGBOOST.