Baseball Hacks

Compare Teams and Players with Lattices

Plot histograms of batting averages for each team in just a few lines of code.

Lattices are a powerful technique for plotting lots of different graphs at once. With the Lattice package, you can divide observations into multiple groups (for example, teams, positions, and stadiums) and draw a different plot for each group. You can also use the package to plot any combination of graphs in one shot—say, a scatter plot, histogram, pie chart, and bar chart—but I think that’s a lot less interesting. Lattices are incredibly useful for comparing groups of teams or players. At a glance, you can quickly see how one group is different from another. You can do this in Excel (plot lots of little diagrams for different subsets of players), but it would take you quite a while.

For these examples, I looked at 2003 batting averages by team. I was curious how the distribution of batting averages differed between teams. Were some teams spread wide apart, with a large difference between best and worst teams? Were other teams packed closely together, with a large number of similarly performing players? Let’s see if there is anything interesting in the data.

In this hack, I examine statistics from 2004, taking only players with more than 250 at bats. (Normally, I like to use 502 plate appearances, which is the number a player needs to qualify for MLB awards. However, there weren’t enough players with qualifying at bats to make these charts interesting, so I reduced ...