Re:Statistical vs. Deep Learning forecasting methods
by MrMan on Thursday December 01, 2022 @12:15PM
(#33819287)
why are middle-ground (but SOTA) techniques like guassian processes and GBM regression not in this comparo
[ Parent
| on HN ]
Re:Statistical vs. Deep Learning forecasting methods
by maxmc on Thursday December 01, 2022 @12:53PM
(#33819851)
A lot of M3 datasets we use are high-frequency, with large seasonal inputs. Considering Gaussian Processes (GP) complexity is O(N^3), a careful study of their performance would be challenging.
Also... I'm not aware of any efficient GP Python implementations.
Re:Statistical vs. Deep Learning forecasting methods
by vladf on Thursday December 01, 2022 @01:16PM
(#33820217)
GPs over time series can leverage low-dimensional index sets for O(N lg N) fitting and inference. This can be done by interpolating the inputs onto a regular grid which admits Toeplitz kernels. See https://arxiv.org/abs/1503.01057.
Re:Statistical vs. Deep Learning forecasting methods
by thanatropism on Thursday December 01, 2022 @12:59PM
(#33819960)
Just write your GP model in Pyro or something like that.