Subscribe Bookmark RSS Feed
tony_cooper1

Staff

Joined:

Sep 29, 2014

GenReg and Change Points in Baseball History

During an era, i.e, a block of time, the rules seem fairly constant. Between eras, the rules may change. This was an issue when considering modeling a disease over time; the rules might change when an effective treatment is released. Less critically, could I empirically replicate the era's in baseball. A paper at the Discovery Summit suggested using GenReg (https://community.jmp.com/docs/DOC-7736). A dataset,modernera.jmp, is attached.

11326_Screen Shot 2016-04-14 at 11.21.43 AM.png11348_Screen Shot 2016-04-14 at 11.27.24 AM.png
It requires creating many dummy variables. For each year, create a column with a formula "less than that year". The formula for a column "1926" is shown. The fist tab In the script automates this.The column "yearID" contained the time series. Check the box to start GenReg, this can easily be achieved with the existing options in "Fit Model". RPG is the Y since we are interested in change points in runs per game. Saving the prediction formula shows all the information you need for change points. However, I wanted to create labels for the empirical periods, which is the reason for the second tab created by running the script.
11349_Screen Shot 2016-04-14 at 11.35.03 AM.png11350_Screen Shot 2016-04-14 at 11.35.51 AM.png
Article Labels
Article Tags
Contributors