r values estimated by SS model... not the right sign for my liking
Closed this issue · 2 comments
For a few of the sites, the mean r value is negative (it should be positive both for the SS and FM models). Site 1 seems to do the best job with r, but site 3 really keeps thinking we should be heading back towards summer all the time.
This must be a problem with our inputs to the SS model. Yikes?!
I'd also note that your prior probability of a negative r should be zero.
On Dec 17, 2013, at 4:51 PM, dgianotti notifications@github.com wrote:
For a few of the sites, the mean r value is negative (it should be positive both for the SS and FM models). Site 1 seems to do the best job with r, but site 3 really keeps thinking we should be heading back towards summer all the time.
This must be a problem with our inputs to the SS model. Yikes?!
—
Reply to this email directly or view it on GitHub.
Prior on r set to be exponentially-distributed. Improves many things.