empty forecast of deepVAR even though the model successfully trained

[ShirleyMgit]

1. I am training deepVAR model, I succeed to train the model but it fails to predict. using make_evaluation_predictions() results in empty list (after doing list(forcesat_it)). can someone please explain why is it happening?
2. In a differennt scenario I can get the forcasts list but the MultivariateEvaluator(quantiles=[0.1, 0.5, 0.9]) does. not work, I get: stopIteration, why?
3. I would be happy to know how can I get the predicted probability distribution parameters

[abdulfatir]

Can you post a snippet of how you're evaluating the model?

[ShirleyMgit]

forecast_it, ts_it = make_evaluation_predictions(test_tf, predictor, num_samples=100)
print("EVALUATION is complete SUCCESS")
forecasts = list(forecast_it)
tss = list(ts_it)
print("after the list convertor: EVALUATION is complete SUCCESS")

evaluator = MultivariateEvaluator(quantiles=[0.1, 0.5, 0.9])
agg_metrics, item_metrics = evaluator(ts_it, forecast_it, num_series=len(test_tf))
print("METRICS retrieved SUCCESS")

[abdulfatir]

I see a potential issue with your code which could be the reason for your problem. You're consuming iterators using list() and then passing the iterators again to the evaluator. These are single use iterators. Either pass the lists to the evaluator as

agg_metrics, item_metrics = evaluator(tss, forecasts, num_series=len(test_tf))

or do not convert the iterators to list first.

[ShirleyMgit]

thanks, is it possible to include the "feat_static_cat" in deepVAR, from some reason I can train the model but than it fails again in transforming the forecasts into list:
forecast_it, ts_it = make_evaluation_predictions(test_tf, predictor, num_samples=100)
print("EVALUATION is complete SUCCESS")
forecasts = list(forecast_it)
tss = list(ts_it)