Data Science in Government

Open data is government is a fun concept. People like it because it implies that all the work that’s being done behind a grey concrete wall will be exposed. But, the truth is, a lot of times there are two issues that a lot of people don’t talk about. So, let’s talk about them.

The first issue is the quality of the data. And it’s a big issue. Just because a gov’t tracks something doesn’t mean that they track it well or correctly. Case in point: My Parks department started tracking fixed assets last spring. Our GIS department built an awesome app that geotagged assets, allowed the foreman to take pictures and then rate the asset on a scale of 1-5 (1 being worst, 5 being best).

This sounds like great information to share with the public. Until you audit it. Then, we quickly realized that there were issues:

1. The ratings were all over the place. 1 was supposed to in such bad shape that it couldn’t be repaired and had to be replaced. 5 was supposed to be perfect. You should be able to tell from my tone that this wasn’t always true.

2. Assets were missing. Things we knew to be in the Parks weren’t showing up.

3. The scale was confusing. We hadn’t done a proper job training our folks on what each number was. Plus, there wasn’t a clear distinction between a 4 (good shape) and a 5 (like new) from an operational standpoint.

This leads to the second point: Just because something seems like it should be shared publicly doesn’t mean it’s ready to be shared. The moral of our story is we made changes, are re-inventorying and will share info later. I look forward to your feedback on the ~4K assets then.

Lastly, for context. The issue with the tool was training, not the tool. That's a concept in software and tech that people don't get, but is usually 100% true. GIS built a great tool that we didn't use particularly well.

We worked with them to create a dataset that we could run some analysis off of and we're using that to target where we're auditing.