Tuesday, March 03, 2015

Little League, Big Data

Last week, I participated in my first Little League draft for my son's baseball team.  This was new territory, as up until now, play has been non-competitive.  This year we will actually have to keep score, and there will be winners and losers.

In preparation for the draft, we had tryouts a few weeks ago where we evaluated the kids on a number of different criteria.  Never have I seen so many scared 7 and 8 year olds march through the cages as dozens of coaches with clipboards watched and recorded their every move.  I camped out and watched them pitch, as from what many veteran coaches told me, the key to keeping the game moving along is the pitcher.

In preparation for the draft, we were sent a couple of key spreadsheets.  The first one had an average rating of all of the kids tryouts assessments, done by the board members.  The second one contained coaches evaluations for some of the players from past seasons. Lots and lots of nothing more than raw data.

Time to fire up APEX.  I created a workspace on my laptop, as I was not sure if we would have WiFi at the draft.  From there, I imported both spreadsheets into tables, and got to work on creating a common key.  Luckily, the combination of first and last name produced no duplicates, so it was pretty easy to link the two tables.  Next, I created a simple IR based on the EVALS table - which was the master.  This report showed all of the tryout scores, and also ranked each player based on the total score.

Upon editing a row in EVALS, I had a second report that showed a summary of the coach's evaluation from prior seasons.  I could also make edits to the EVALS table, such as identify players that I was interested in, players that were already drafted, and any other comments that I wanted to track.

After about 20 minutes of reviewing the data, I noticed something.  I was using data collected while the player was under a lot of stress.  The data set was also small, as each player only got 5 pitches, 5 catches, 5 throws, etc.  The better indicator as to a player's talents was in the coach's evaluations, as that represents an entire season of interaction with the player, not just a 3-4 minute period.

Based on this, I was quickly able to change my IR on the first page to also include a summary of the coach's evaluations alongside the tryout evaluations.  I sorted my report based on that, and got a very different order.  This was the order that I was going to go with for my picks.

Once the draft started, it was very easy to mark each player as drafted, so that any drafted player would no longer show up in the report.  It was also trivial to toggle the "must draft" column on and off, ensuring that if there were any younger players that I wanted, I could get them in the early rounds before we had to only draft older players.

Each time it was my pick, I already knew which player that I was going to draft.  Meanwhile, the other coaches shuffled stacks of marked up papers and attempted to navigate multiple spreadsheets when it was theirs.  Even the coordinator commented on how I was always ready and kept things moving along.

Unless you're some sort of youth athletics coach that does a draft, this application will likely do you little good.  But the concept can go a long way.  In almost any role in any organization, you likely have data for something scattered across a few different sources or spreadsheets.  This data, when isolated, only paints a blurry part of the whole picture.  But when combined and analyzed, the data can start to tell a better story, as was the case in my draft.

The technical skills required to build this application were also quite minimal.  The bulk of what I used was built-in functionality of the Interactive Report in APEX.  Merging the data and linking the two tables was really the only true technical portion of this, and that's even something that can be done by a novice.

So the next time you have a stack of data that may be somehow related, resist the temptation to use old methods when trying to analyze it.  Get it into the database, merge it as best you can, and let APEX do the rest.

No comments: