ExtremeValueAnalysis - Max Data Points Issue

[DAGerretzen]

Hi AI team,

Thanks for this app - in general works great but we have come across a bug where it fails if given an input timeseries of over 500000 data points. After digging a little deeper looks to be due to using getTimeSeriesData. Was easy enough get around this by switching to getTimeSeriesCorrectedData and making a few small changes elsewhere.

No changes necessary from our side but thought it best just to make you aware of what we found.

Cheers

Darren Gerretzen (BOPRC)

[TheRealDougSchmidt]

Hi Darren.

Yeah, the GetTimeSeriesData API has a limit of 500K points in the master time-series. I've just confirmed isn't publicly documented (it should have been, but it isn't). The idea is to avoid causing too much memory pressure on the server when it attempts to interpolate the values of all the secondary time-series.

But the ExtremeValueAnalsys app doesn't even pull more than 1 time-series at once, so it doesn't really need to use this API.

My hope is that the R and Python wrappers can eventually mask this constraint, and fetch time-aligned data from the server in series of batch requests, and then just stitch all the data back into a single data frame. That magic plumbing doesn't exist yet, as you've discovered.