Hi,
I was wondering whether we could include multi-scale time series as features: say aggregate daily data with hourly data for example. I guess that would require a multi-branch (multi-input) model? In some context like e.g. financial data that may improve forecast.
Where and how would you do the aggregation and how would you deal with the problem that the features of different aggregations (say hourly and daily) are of different length.
Hi, I was wondering whether we could include multi-scale time series as features: say aggregate daily data with hourly data for example. I guess that would require a multi-branch (multi-input) model? In some context like e.g. financial data that may improve forecast.
What do you guys think ?