Hey Hadjmbarek Nadia ,
Thanks for reaching out on Microsoft Q&A.
Good question! So, you've got target_lags = [0] in your AutoML experiment. This means the model's only looking at the current value to predict the future. It's kind of like trying to predict tomorrow's weather based solely on today's conditions, without considering past weather patterns.
If you want the model to learn from the past, you can either manually set specific lags (like [1, 2, 3] to look at the previous 1, 2, and 3 time steps) or let AutoML choose the best lags for you by setting target_lags = "auto".
Hope this helps!
If you found this answer helpful, please upvote and mark it as accepted to close the thread. Thanks!