search

awslabs / gluonts

Probabilistic time series modeling in Python
https://ts.gluon.ai
Apache License 2.0
4.54k stars 746 forks source link

How to run forecasts in parallel #705

Open rt3722 opened 4 years ago

rt3722 commented 4 years ago

Hi,

When we have large number of time series to forecast like say above 150K then below steps take lot of time.

forecasts = list(forecast_it)
tss = list(ts_it)

Does anyone know to run the above steps in parallel lets say using joblib parallel or multiprocessing

jaheba commented 4 years ago

Creating predictions in parallel should be easily doable.

If you are just interested in getting predictions, you can distribute the input-data across multiple instances of estimators, by calling estimator.predict(...) on them (given you use a library which allows you to do these things).