Monday, July 3, 2017

Chainsaw carving: part 3

Every data analysis has its ups and downs. My analysis of oyster larvae behavior has been mostly up recently. But after about 6 months of work, the analysis still isn't quite finished.

In a meeting, one of my collaborators on the project pointed out some nonsensical numbers in the dataset. She asked me to take another look, and I discovered that my code was miscounting the number of larvae entering each experimental flask. You'd think that counting the larvae would be the simplest task, right? It's surprisingly difficult, because no single set of parameters applies to all experiments without error. There was no way to automatically count larvae.

So I started over and did it manually.

Friends, this is just how it goes. In fact, I don't think I can name a single scientific paper I've published for which I haven't had to start over at least once. Every analysis requires at least one trip back to Square One.

The good news is that now the data look a lot cleaner, and we're mere steps away from finishing the analysis. To continue my chainsaw metaphor, I haven't scrapped my sculpture; I've hollowed it out. I have used my chainsaw to carve large swaths of wood from the inner regions of my creation. Wood chips flying, I have sliced and dug and carved until the interior was empty. It began as a tree trunk, and my sculpture is almost done.

No comments:

Post a Comment