search

ngruver / llmtime

https://arxiv.org/abs/2310.07820
MIT License
628 stars 139 forks source link

Size of the test set of the Informer datasets #20

Open itdevwu opened 6 months ago

itdevwu commented 6 months ago

Hi Nate!

Just scanned through you marvelous work. I found that the precomputed output of Autoformer on the Informer datasets are substantially smaller than the original test set. As you mentioned in your paper that the test set has been narrowed, but what is the actual size of the test set?

ngruver commented 6 months ago

HI Astor,

Our test set has one chunk per covariate and they are predicted independently. For the datasets with a small number of covariates, you could probably run evaluation with the original test set, but with many covariates it is intractable, as the number of series to predict will be (number of covariates) x (number of temporal chunks).

Nate