commfish / seak_sablefish

NSEI sablefish stock assessment
8 stars 5 forks source link

Oddity in proportion mature #24

Closed jysullivan closed 5 years ago

jysullivan commented 5 years ago

There appears to be a dip in the proportion mature at age-14. This function is from a logistic regression and should be a smooth line. Need to figure out if this is it a plotting artifact or what's going on.

image

jysullivan commented 5 years ago

I don't understand this one. Here is the underlying data:

image

There's nothing here that explains that dip for age-14, though I agree it's annoying.


To clarify these predictions are not directly from a logistic regression. When I compared age-based to length-based maturity curves, the length-based curve had better fit by AIC. I also looked at age by year and length by year combinations. The figure above represents a length-based logistic regression for maturity that was translated to age using the fitted curve for a length-based vonB from the same years.

Here is the difference between the age-based (red) and the length-based (blue) that has been translated to age. Light lines are annual mean predictions, dark lines are the means. image

The AIC comparison: image

ben-williams commented 5 years ago

That makes sense - I hadn't realized that it was back transformed age/length model fit. I wouldn't worry about it then