This message was deleted.
# neural-forecasts
Slackbot
09/19/2023, 9:45 PMThis message was deleted.
c
Cristian (Nixtla)
09/19/2023, 10:35 PMHi @Alex Wang, this type of normalization is increasingly common in time series. We have seen consistent good results with this approach, in particular in settings with large scale variation within and across time series.
Cristian (Nixtla)
09/19/2023, 10:36 PMHowever, we know that it has some limitations (for instance, loss of information). We are working on adding a separate pre-processing step that performs normalization over the complete time series
a
Alex Wang
09/19/2023, 10:36 PMGotcha this is incredibly helpful! Thank you! Is there a reference that evaluates this type of temporal norm?
c
Cristian (Nixtla)
09/19/2023, 10:38 PMFor example, we just saw that our normalization can drop vital information in healthcare settings (your area?). Normalizing in a short window removes the information of the medication dosage, since it transform it to essentially a dummy (0-1)
❤️ 1
a
Alex Wang
09/19/2023, 10:39 PMExactly, that's what I was worried about since sometimes the baseline level is very important information
c
Cristian (Nixtla)
09/19/2023, 10:39 PMWe discuss TemporalNormalization here: https://arxiv.org/abs/2305.07089
Cristian (Nixtla)
09/19/2023, 10:40 PMOne alternative is to simply normalize the data before, and set
scaler_type=NoneCristian (Nixtla)
09/19/2023, 10:41 PMFinally, some models such as the NHITS and NBEATS are extremely robust to the scale. We have seen amazing results without any normalization
a
Alex Wang
09/19/2023, 10:43 PMthanks a lot for these references and pointers 🙂
16 Views