Measuring Video Game Engagement through Social Media

This Semester was undeniably the most data heavy. Of the classes I was taking, the only non-math related course was Software Lab. Even then, our final project had a huge data scraping requirement.

We were a team of 5, shaping the gaming industry forever. The concept was there: display as much information as we can on the different consoles so that a person new to gaming could make an educated decision on which one to buy. The problem was what were we going ACTUALLY do? What information did we need to find in the first place?

I decided I wanted to do something innovative. Rather than use user ratings and critic ratings which aren't always accurate, why don't we measure a game by how much it appears in Social Media?

Honestly, this project was pretty huge and time intensive. I couldn't have focused on the data without the others doing the web development, apps, and back end. I'm grateful that we were able to pull through and deliver this social media engagement feature.

In theory, it's pretty simple. Find the number of times a game title appears in social media posts with X filter over Y number of days. For each title, put each social media metric in its own column. Do this for all of the games. Normalize the columns and perform basic stats to see how video games are appearing across the social media landscape.

Implementation was its own beast. I needed to take a list of games, clean it, use APIs to query the social media platforms, clean the results, analyze the data, etc. Because of the sheer amount of games, it took a really long time. It didn't stop there though. My teammates needed to store the data, filter the data, and retrieve the data which was a process in and of itself.

At the end of the day, the results were interesting to see. The games at the top were niche market. They had massive followings from loyal fans that only cared about those games and their respective communities. The more complex games with communities and discussions fared better for obvious reasons. Also, game scores were exponentially decreasing in that many games had low engagement but the number of games that had "high" scores rapidly decreased as the score threshold increased.

All in all, it was a great experience working with an unconventional data set.