8:00 – 4:30 SR
- Showed Lenny the new charts. He’ll be out tomorrow, so I’ll probably deploy the test version on Monday. We then had a long talk about the best way to statistically model the data in the first chart. It really sounds like Bayes might be the way to do this. Which is awesomely cool, because my schoolwork and jobwork are perfectly aligned for the weekend. Gonna charge me some hours figuring out Bayes.
- Building the dataFormat and dataProvider Objects. Since Lenny wants future performance, I’m pulling in the Apache commons math jar for linear regression. I think I want OLSMultipleLinearRegression
- Added SimpleLinearRegression to JavaUtils that wraps OLSMultipleLinearRegression and hides some complexity.that is not intuitively obvious documentation. Glad I found this example. Added to Subversion. I’ll probably need to add a method that gets the error so that I can add that to the future error.
- More JSON tomorrow, then on to Bayes. Did get this book…