I’m moving to Houston next month, and I’m working furiously to finish up about 4 papers before I leave. Yesterday was a good day because two things worked absolutely perfectly.
I’m calculating the likelihoods of some of our observations and I wrote a specific routine to do it for a first-order model. I then wrote a generic routine to calculated likelihoods for higher-order models. When I used the generic routine to calculate the likelihood for a first-order model, I got the same result as my specialized routine. Yatah!
From previous analysis using partial autocorrelations, we determined that a third-order model should explain our data the best. When I compared our models to see which one was most parsimonious (using AIC), the third-order model again came out on top. Yatah2!
test
test
test
test
test
test
Testing comments.
Update