Affiliate Status Identification using Twitch Streaming Data

Introduction

This project provides a comprehensive analysis of Twitch streaming data, focusing on the graph analysis, feature engineering, and predictive modeling using Graph Convolutional Networks (GCN). By dissecting user engagement metrics, account features, and network structure, we gain insights into factors influencing content creators’ success and the underlying social network dynamics. Additionally, we also highlight the use of predictive modeling in identifying Twitch affiliate status, which is crucial for targeted advertising and facilitating collaborations, thereby enhancing the monetization strategies for creators.

Dataset

Stanford Network Analysis Project (SNAP) - Twitch Social Networks (Spring 2018)

Contents

• Jupyter Notebook

• Project Report