Project Background

This project is a group effort collaboration for Syracuse University’s Marketing Analytics coursework. The project explores multiple data sources and data collection methods to attempt to organize streaming and gaming popularity data to be analyzed to observe the relationships between game success, game interest and game streaming metrics.

Overview

The gaming community is unique in behavior, but is more unique in the potential to record, predict and model varying behavioral patterns within the community. Top streaming game data was taken from https://sullygnome.com/, while https://www.vgchartz.com/ supplied game sales/success data, and Google Trends was used as a proxy for game interest data. Some preliminary survey data was taken from the undergraduate Marketing Analytics class at Syracuse University’s Whitman School of Management.

Data was mined and engineered using a custom web crawler application. The data was then used to construct simple linear regression. While the best predictive models performed poorly, primarily as an artifact of the sparsity and inconsistency of data sources, although some correlative information was mined. Below, we see the correlations between Twitch platform metrics across games is virtually unassociated with game interest data.

This unique disconnect is reflected in findings across the Twitch community as a subset of the gaming community. Though there is some interplay between the popularity of a game and how the game performs on Twitch, it seems that this is a case by case basis and would require more data to look into.

Link to GitHub Repo