Hello i'm trying to identify the different time serie forecasting methods i could use according to my specific problem:
I would like to perform:
- A Multi-Step
- Multi-Variate (some covariates with seasonality)
- Forecasting problem
- without knowing previous values of the target (= no history of the target available prior to prediction point)
- On a target with trend but no seasonality
I'm considering LSTM/attention models but would like to know if a simplier model would be useful. I think the auto-regression part of SARIMAX implies that previous target values are needed for training the model. Any others advices? Thanks!
