ENHANCING MUSIC DISCOVERY: A PREDICTIVE MODEL FOR A SONG REPETITION IN MUSIC STREAMING SERVICES

Abstract

The music industry has seen a remarkable transformation over the past few decades, with streaming platforms and digital analytics providing unprecedented access to detailed song performance metrics. Statistical analyses of song popularity reveal notable trends, such as the exponential growth of streaming numbers from 2010 to 2023, where average monthly streams for top-charting tracks increased by over 300%. Additionally, factors such as genre diversity, artist collaborations, and seasonal trends have shown significant impacts on song popularity over the years. Traditionally, determining song popularity relied on manual approaches such as surveys, radio airplay counts, and subjective assessments by music critics. These methods often encountered challenges including limited sample sizes, delayed data collection, and inherent biases. Consequently, they provided a partial and often skewed view of song performance, hindering accurate trend analysis and prediction. Machine Learning (ML) offers a promising solution to these limitations by enabling the analysis of vast amounts of data from various sources, including streaming metrics, social media sentiment, and historical performance trends. ML algorithms can uncover complex patterns and correlations that manual methods might miss, leading to more accurate predictions of song popularity and trends. By leveraging advanced techniques such as neural networks and natural language processing, ML models can provide deeper insights and real-time assessments, transforming the way the music industry understands and anticipates audience preferences