Pop Music Trend and Image Analysis Based on Big Data Technology.

Area of Science:

• Music Information Retrieval
• Data Science
• Computational Musicology

Background:

• The music industry increasingly relies on data-driven insights for artistic creation and trend analysis.
• Understanding future music trends is crucial for artists to create relevant and successful works.

Purpose of the Study:

• To evaluate and compare the effectiveness of different algorithms for predicting music trends.
• To identify the most accurate model for analyzing music data and forecasting future popularity.

Main Methods:

• Introduction to music pop trend analysis theory and big data mining technologies.
• Implementation and comparison of Autoregressive Integrated Moving (ARIM), random forest, and Long-Term and Short-Term Memory (LSTM) algorithms.
• Analysis of music data focusing on collection, download, and playback metrics.

Main Results:

• The Long-Term and Short-Term Memory (LSTM) model demonstrated strong predictive capabilities for music playback times.
• The Autoregressive Integrated Moving (ARIM) model showed minimal data gaps between predicted and actual playback times, indicating an allowable error range.
• While LSTM excels, it struggles with songs exhibiting significant data fluctuations; ARIM provides a more stable, albeit less precise, prediction for such cases.

Conclusions:

• The Long-Term and Short-Term Memory (LSTM) algorithm provides more accurate music trend predictions compared to Autoregressive Integrated Moving (ARIM) and random forest algorithms.
• These findings can guide singers in aligning song creation with current and future music trends.
• The research promotes a more data-informed and modern approach to traditional music creation.