Hello, I'm using the HHITS model pre-trained with M4 data, and trying to adjust it to my data. However, the M4 data is monthly and my data is daily, so when I try to adjust this pre-trained model for my daily data, it is coming out monthly. Is there any way to fix this? Does the 'predict' method have any parameters to adjust this? My code is in this file:

You can override the freq attribute of the NeuralForecast object, e.g.

nf.freq = 'D'

Ok, thanks!!

Hi @Betina Corazza Ferrão. Are you specifically exploring cross-frequency transfer learning? We have seen that this tends to perform poorly, given the different lengths of recurrent patterns

Ah, okay, thank you for the feedback!