- With enough at-bats, might player performance "regress to the mean" in kickball like it does in baseball?
- If so, about how many at-bats should a player take before their talent can be fairly assessed?
- Is it fair to compare the guys to the gals?
- Which players show up to games most reliably?
- How awesome do we look while wearing patriotic attire?
The Data: At the end of the each game I take a picture of the scoresheet. Then, following a trip to the bar for beer, pizza and chicken wings I manually translate the scoresheet into a custom schema (Excel spreadsheet) with one record for every player At-Bat. This schema holds information about Batters, Gender, Hit Types, number of outs and runners during the current at-bat, and even other information about time of day, location, weather conditions, and the weekly on-field beverage. It should be mentioned here that designing and populating the schema was a non-trivial effort; this is a very important but often overlooked precursor to effective use of Tableau.
The Visualization: I quickly and easily connected Tableau to my custom Excel schema and created a Tableau Data Extract (.tde) file. Each week as more data goes in, I simply refresh the extract and the charts adjust automatically. If I rearrange the column order, rename fields, or add new ones Tableau automatically detects these changes and compensates on refresh.
To make the chart, all I had to do was I create calculated fields for OBP, SLG, and OPS. I then dragged OPS onto rows and At-Bats onto columns, whereupon the 'Show Me' card suggested that I use a scatter plot since I was comparing two measures. After dropping Batter Name onto Label and Gender onto the Color I created a trend line, set the intercept to zero and elected to show the confidence bands. Defaulting to a linear trend did not seem to correctly represent the data, so I switched to a logarithmic trend model with a single click. Satisfied, I added annotations displaying the trend equation and the p-value for each gender. Finally, for kicks I sized the data points by Runs Scored and added a background image from after our Tournament the other weekend.
Why Tableau Rules: Within minutes of connecting Tableau to my data I was able to create a very communicative and rich visualization to share with my team. Perhaps more rigorous statistical analysis is likely required to make real conclusions about the data but for the purposes of delivering a message about kickball performance this chart was more than adequate.