https://github.com/nixtla logo
#neural-forecast
Title
# neural-forecast
c

Chris Gervais

11/08/2022, 11:18 AM
btw we’re starting to get some pretty conclusive results re: neuralforecast performance as benchmarked against xgboost, catboost, lightgbm etc. @Valeriy not sure if you’ve had a chance to wrap up your analysis but seems pretty clear at this point the conclusion from https://arxiv.org/abs/2106.03253 is bogus
👍 2
we can definitely attest to a difference in ease of tuning, but the lift against xgboost is significant (at least in our domain) and persists across ~95% of all of our ~1k forecasting tasks
v

Valeriy

11/08/2022, 11:23 AM
XGBoost is all about creating features. Could this be the issue?
c

Chris Gervais

11/08/2022, 11:25 AM
we ran both with and without tsfresh features, same finding held up.
👍 1
v

Valeriy

11/08/2022, 11:26 AM
Nice there is also new one for features could be interesting for team https://github.com/predict-idlab/tsflex
c

Chris Gervais

11/08/2022, 11:26 AM
awesome, will give this a try 👍
v

Valeriy

11/08/2022, 11:29 AM
Did DL win on global models (multi series) or univariate as well?
c

Chris Gervais

11/08/2022, 11:46 AM
yah for clarity this is all for multioutput single series using exogenous variables
👍 1