Apr 23 2007
Research Update and General Grumblings
As I’ve written about before, I’m using principle components analysis (PDF) in an attempt to distinguish two processes that occur in stratocumulus clouds. The problem - or so I’ve been told - is that PCA will separate covariance into orthogonal basis vectors regardless if there are two processes or not. Therefore, when attempting to interprete, it’s important to make sure that the results tell you exactly what you think they tell you. In my case, it’s fairly easy to do: I just look at a different day. Happily, when I did, I got the same results as before… kinda… but I cheated. Okay, cheated isn’t the right word. But when I looked at day x+1, the results were slightly different than day x. Argh. When I looked at day x+2 though, I was in heaven. Exactly the same as day x. I rushed to find the advisor, but he had already left for the weekend - and who could blame him, it was 7:30pm. I decided since he was gone, that was a good excuse for me to leave too, though I probably should have stayed and looked at some other stuff. Add to that that I’m terribly unproductive on Monday mornings. Is anyone productive on Monday morning? Maybe the weekend should be extended until noon on Monday. I’d gladly give up Friday nights so I didn’t have to come into work on Monday.
(No Ratings Yet)
Loading ...Related Posts:
One Response to “Research Update and General Grumblings”
To reduce spam, comments are automatically closed 30 days after the last comment. If you would like to comment on any closed thread, please use the contact form at the top of this page.
If you have already hypothesized the processes, why use PCA? Just do a regression.
[Response: The data are incredibly noisy, and we need some was to distinguish the signal from the noise. We're now looking at other techniques besides PCA.]