Treatment of Individual errors of measurements


I am trying to use individual errors of my inputs into the Stan model. I haven’t find a way to treat them individually. The main problem here is that each entry has it own individual error.
Let’s say I have a “Y” I want to predict from an entry of values “X” of 1000 rows. Each X entry has a different error, so my question is it possible to treat them individually in the pystan model?


Hey l, yes you can treat them individually quite easily . Have a look at the measurement error section in the handbook as a general overview.


yourvalue ~ normal(x, yourerror)

Where x is a parameter indicating the mean you don’t know which you can then use in further calculations.

Ok I think I got it based on the handbook and your reply.

When you said “yourvalue” those are all my X entries right? that I will use to predict my “Y”.

Is there a reason why to peak normal, or it will vary in regards of the type of error?

Thanks ,


Random additive error is Gaussian, but if you’ve some other error model then you should use that instead.