3 ms·
Let's start that AIC is misused here as it is valid only when you have a likelihood function for the data. The mentioned scatter plots do not have any (or have
by AstralStorm 8y ago
Let's start that AIC is misused here as it is valid only when you have a likelihood function for the data. The mentioned scatter plots do not have any (or have something assumed) for prior. Likewise BIC.
The posterior likelihood function has very interesting values for this logistic map fit.
This is on the level of "gotcha". Usually the AIC and BIC are supposed to be negative for a correct model and the for is supposed to have a quantifiable information loss.
- jmalicki 8y agoAIC and BIC are only really valid when comparing nested models anyway.
- gammarator 8y agoThis article at least claims that's not right: https://robjhyndman.com/hyndsight/aic/ https://robjhyndman.com/hyndsight/aic/ (And Wikipedia: https://en.m.wikipedia.org/wiki/Akaike_information_criterion#How_to_apply_AIC_in_practice https://en.m.wikipedia.org/wiki/Akaike_information_criterion...) Do you have a reference? Or were you thinking of the Likelihood Ratio Test?
- jmalicki 8y agoOkay reading, it sounds like before Burnham and Anderson (2002) it was only known to be valid for nested models, but they were able to derive it from more fundamental assumptions to remove that restriction. https://en.wikipedia.org/wiki/Akaike_information_criterion#History https://en.wikipedia.org/wiki/Akaike_information_criterion#H... I wasn't aware of that update until now (despite being 16 years old), thank you!!