R freeze during predict

[vivshume]

Hi ,
when I am running my Model for 15k+ prediction, My model suddenly stops in middle and it still shows processing. The model never moves forward.
Can you Please let me know the Fix ?
Thanks
Vivek

[TheGeoLawton]

I wonder if this might be an internal memory issue. R stores large objects in memory, and when the machine fills up, the processing stops. If that's the case, It is not specific to Prophet, it is an issue with R itself.

[bletham]

This is probably related to #215. Basically, in prediction we make a really big matrix that is (number of prediction points)x(uncertainty_samples), where uncertainty_samples here is an input to Prophet() for how many samples to use to estimate trend uncertainty. 15,000x1000 definitely sounds big enough to cause memory issues.

There are a few things you can do to get around this. If you don't care about uncertainty, then set uncertainty_samples=1 when you create the Prophet object. Alternatively, just break your predict dates into chunks of maybe a 1000 and call predict separately on each chunk, and then concat the output dataframes from each call. Does that make sense?

[bletham]

I'm hoping these suggestions resolved the issue, otherwise feel free to re-open.