Music Development Network Analysis

This repository contains the code and data for our project on analyzing the influence relationship between musicians using neural networks.

Project Structure

• Data: this folder should focus on 2 datasets:

  • full_music_data.csv: This dataset includes 16 variable entries for 98,340 songs. These variables include attributes like danceability, tempo, loudness, key, and artist information (name, id, etc.).
  • influence_data.csv: This dataset covers musical features for 5,854 artists spanning the last 90 years. It provides both influencing artists and their respective followers.

• 1.music_data_manipulation.ipynb: This Jupyter notebook is used to manipulate and preprocess the music data. Key steps include matching musician IDs, cleaning the data, and preparing it for feature engineering and model training.

• 2.musician_sequential_input_gen.ipynb: This notebook is crucial for feature engineering. It generates independent variables by selecting 20 songs per artist in chronological order, creating a sequential representation of each artist's music. These sequential inputs are vital for capturing the temporal and stylistic progression of an artist’s work, enabling the model to effectively assess influence patterns.

• 3.musician_influence.ipynb: This notebook is used to train the final model. The model is based on TensorFlow and aims to analyze and predict the influence relationships between different musicians.

• influence_musicians8570.h5: This file contains the final model weights, trained to evaluate the influence relations.

• IEOR242-Project-Group6-Statistical Analysis on Music Influence Relation with Neural Network.pdf: The comprehensive report detailing our analysis, methodologies, results, and conclusions.

Dependencies

To run the code in this repository, you will need the following Python packages:

• Python 3.7 or higher
• tensorflow
• Other dependencies as listed in individual scripts