Litcius/Paper detail

Predicting a Hit Song with Machine Learning: Is there an apriori secret formula?

Agha Haider Raza, Krishnadas Nanath

202026 citationsDOI

Abstract

Thought to be an ever-changing art form, music has been a form of recreational entertainment for ages. The music industry is constantly making efforts for songs to be a hit and earn considerable revenues. It could be an interesting exercise to predict a song making it to top charts from a mathematical perspective. While several studies have looked into factors after a song is released, this research looks at apriori parameters of a song to predict the success of a song. Data sources available from multiple platforms are combined to create a dataset that has technical parameters of a song and sentimental analysis of the lyrics. Four machine learning algorithms (Logistic Regression, Decision Trees, Naïve Bayes and Random Forests) to answer the question-Is there a magical formula for the prediction of hit songs? It was found that there are elements beyond technical data points that could predict a song being hit or not. This paper takes a stand that music prediction is yet not a data science activity.

Topics & Concepts

Random forestLyricsComputer scienceMachine learningNaive Bayes classifierA priori and a posterioriEntertainmentArtificial intelligencePerspective (graphical)Decision treePredictive modellingData scienceSupport vector machineArtVisual artsEpistemologyLiteraturePhilosophyMusic and Audio ProcessingMusic Technology and Sound StudiesGenerative Adversarial Networks and Image Synthesis
Predicting a Hit Song with Machine Learning: Is there an apriori secret formula? | Litcius