Machine Learning Classification Techniques Applied in modern-day Music Recommendation Systems

Abstract

Music streaming platforms have increased in demand over recent years.Due to the covid-19 pandemic and lockdowns, there has been a substantial increasein users in music streaming platforms, where the music streaming market grew by $7.47 billion. Hence, the importance of recommendation systems in these platforms. The goal of this research paper is to address and provide insights into the behind the scenes to recommendation systems. To have the most accurate results, this paper will be based on pre-existing GTZAN audio samples as datasets for different classifiers and compare the results of accuracy and consistency of the model’s outcome for each genre. Confusion matricesare the finest way to determine results from each classifier. To identify limitations, advantages, and whether or not each classifier provides the most ideal outcomes that satisfy user demands. This process will include processes such as the sample input, feature extraction, and finally, put through classifiers.