If you have multiple time series data for a given problem (e.g predicting house prices and data is available per city). Per city there is a list of features and the target feature. If you want to train one LSTM of all the cities together, how you would approach that?

I was thinking of using a stateless LSTM architecture where I organize my input in such a way that each batch represent a time series of a city. If that approach would work, are there more things I need to account for?

What about making additional features with distance to other cities, thoughts on that?



You must log in to answer this question.