For years, promotional departments in MLB teams have been using bobblehead giveaways to squeeze extra attendance to the games because fans really love these little men. However, since there many so factors affect someone’s decision to go to the ballparks (to name a few: weather, time, winning percentage of the home team, opponents) and these variables vary across MLB teams, truly understanding the effect of bobblehead on attendance is important to each MLB team. In this analysis, we try to tackle this problem by performing the analysis on the dataset about attendance in 2012 season. This is panel data meaning that it contains cross-sectional and time series data at the same time. We will use different approach with different level of aggregation to really understand the effect of bobblehead given the heterogeneity between teams in MLB.
This project is inspired by one of the case studies in the book “Marketing Data Science: Modeling Techniques in Predictive Analytics with R and Python.” by Thomas Miller. The author provided the dataset and problem statement.
The analysis contains 3 main parts:
- Exploratory Data Analysis: understanding the dataset
- Model Selection & Comparison: We will compare between 3 approaches: aggregation, grouping and multi-level with level of aggregation reducing to clearly understand the effect of bobblehead on game attendance of different teams
- Recommendation - value added recommendations to marketers in MLBs
The folder contains:
- Dataset of bobbleheads (bobbleheads.csv, city.csv)
- EDA and Modeling (main.ipynb)
- Final Report (Report.docx)
Check out my Medium for more projects: https://medium.com/@fumanguyen.ymc