Okay @ckrapu I got it ! Now it goes but, when I’m running the model:
I realized that the linear formula have very strange weights!
Variable: Intercept Mean weight in model: 115.0491 Variable: DIA Mean weight in model: 339.0715 Variable: CANTIDAD Mean weight in model: -1446.7739 'prob_total_tipo_movimiento = 115.05 * Intercept + 339.07 * DIA + -1446.77 * CANTIDAD'
(basically I want to calculate, later, the probability between 0 and 1 that a bank movement is a supermarket movement taking into account the day and the amount, which I have specified in my dataset)
So, this is my last question about this:
when I try to sample, I get lot of divergences:
I tried to raise the target_accepted range, increase tune but that didn’t succed, I get divergences anyway.
I was thinking and maybe it’s the way I’m defining the linear model in model_fullsyntax.
Could you tell me what I might be doing wrong about the model that makes it diverge?
EDIT (if it helps):
my dataset is → cuenta_normal_añoentero.csv (2.5 KB)
and I make these transformations on the ‘features’ that I consider important to the model:
Thankyou so much !
Sorry if that’s a rookie question, I’m trying to learn everything about pymc3 this quarantine.