Thursday, October 18, 2007

Linear Regression and Matrices

So I know a good amount of math, more than the average guy on the street, but I love to learn new things. This week I passed out copies to the book Supercrunchers to my two kids in math club, and to get the ball started we went over how linear regression is calculated. I had no idea, and I found the following link that does a really great job, but it does almost the whole thing with matrices. I had never seen this before, and I was basically stunned. One nice thing about it is that it gives a good application of the matrix transpose. The fact the calculation of the linear regression numbers can be written as a fairly short matrix equations is shocking to me. This of course reminds me of determinants, which continue to amaze with their power and breadth of application.

No comments: