This is not an issue reporting, but rather an ask for suggestions, as there are not many longevity modelers on other major sites like StackOverflow.
I am new to longevity modeling and saw that Andres is out of the office until Jan 15th, but I just want to ask longevity modelers in general, that is it possible to obtain good training errors in general?
Following the StMoMo Vignette and using data for USA, and training the APC model on data from 1977 - 2014 for ages 40-80, I have the following percent errors:
. The errors are highlighted according whether they are positive or negative. I know that percent error may not be a good metrics and MSE might be better, but I chose percent error primarily for its interpretability (please bear with me).
I am wondering if this is normal by any means, and if there are ways to improve training accuracy.
This is not an issue reporting, but rather an ask for suggestions, as there are not many longevity modelers on other major sites like StackOverflow.
I am new to longevity modeling and saw that Andres is out of the office until Jan 15th, but I just want to ask longevity modelers in general, that is it possible to obtain good training errors in general?
Following the StMoMo Vignette and using data for USA, and training the APC model on data from 1977 - 2014 for ages 40-80, I have the following percent errors:
. The errors are highlighted according whether they are positive or negative. I know that percent error may not be a good metrics and MSE might be better, but I chose percent error primarily for its interpretability (please bear with me).
I am wondering if this is normal by any means, and if there are ways to improve training accuracy.