The project that made me realize what I enjoy most about analytics. As a lifelong Astros fan, I've always wondered what actually drives attendance. Winning games, the day of the week, or something else entirely?
What actually drives attendance at Minute Maid Park? The Astros finished the 2023 season with 90 wins, another division title, and a deep postseason run. Most people would assume winning naturally leads to larger crowds. I wanted to find out if that assumption was actually true. So I tested whether game outcomes and game timing could explain attendance across an entire season, rather than just accepting conventional wisdom.
Using publicly available attendance data from all 81 Astros home games in 2023, I built a multiple linear regression model with two independent variables, game outcome (win or loss) and game timing (day or night), to measure which factors had a meaningful impact and which did not:
Attendance = 37,549.29 − 984.91 × Win_dv + 2,052.75 × Day_dv
The results surprised me. Day games consistently attracted more fans and were the only statistically significant predictor in the model. Winning games, despite being the assumption most people would make, was not statistically significant. That was my favorite part of the project. The data challenged what I thought I already knew.
This project reinforced something I still believe today: the value of analytics isn't confirming what we already expect. It's uncovering something we didn't know before. Whether the question involves sports, consumer behavior, retail performance, or business strategy, I enjoy starting with curiosity, testing ideas with data, and translating the results into decisions people can confidently act on.
Like every good analysis, this project generated more questions than answers. Future models could include additional variables such as opponent quality, promotions, weather, ticket prices, giveaways, holidays, and rivalry games to better understand what truly drives attendance. That's what I enjoy most about analytics. Every answer creates another question worth exploring.