Music Recommendation System Using Big Data and Machine Learning

Abstract

In comparison to earlier times, there is now a wealth of music available thanks to the development of music streaming services that can be used from mobile devices. Classifying digital music can be timeconsuming and cause fatigue. Obtaining one’s favorite music from the huge music selection has become a problem. The creation of a music recommendation system that can search through music libraries and make song recommendations to users based on their listening habits is therefore useful. With the aid of music recognition, the system can forecast and then suggest songs to its users based on the traits of the music that they have previously listened to. The system collects information, then it reads the data and identifies the category of the music. Classification works towards the goal by comparing the similarities between the music categories and customer preferences. On the whole, results show that accuracy rate achieved for the single music category user is turned out to be higher than multicategory music user with an average accuracy rate for single category being 50.15%, whereas accuracy for feature recommendation in multicategory is higher than single category with the accuracy being around 42.80%. Based on these results, it shows that although the accuracy for recommendation system is not high, it has its own advantages in the start and can optimize itself for better recommendations.