https://github.com/nixtla logo
v

Valeriy

10/06/2022, 3:43 PM
Hi Nixtla team, I am planning to deep dive into library what top 3-5 most performant models you would recommend me to look into @Max (Nixtla) @fede @Kin Gtz. Olivares
m

Max (Nixtla)

10/06/2022, 3:46 PM
Hi Valeriy, hope everything is good :) You mean NeuralForecast or StatsForecast or in general?
👍 1
v

Valeriy

10/06/2022, 3:51 PM
3-5 across the whole of Nixtla maybe at least 1-2 from neural and stats each. Most performant ones in terms of Nixtla team experience.
m

Max (Nixtla)

10/07/2022, 11:32 AM
Sure! From stats: • auto ETS • auto Arima • CES (complex exponential smoothing) neural: • NHITS • NBEATS
🙌 1
🙏 1
Happy to help if something comes up 🙂
BTW, @Valeriy, you might also be interested in playing around with GitHub.com/nixtla/mlforecast
Your feedback would be very appreciated
v

Valeriy

10/11/2022, 9:24 AM
Thanks, @Max (Nixtla), on it, might take some time as quite busy but will reach out. I have something in mind on how to test them all on large-scale data.
m

Max (Nixtla)

10/11/2022, 2:35 PM
That’s amazing, we are particularly interested in scalability and efficiency, so happy to help. We have some good experiences with Ray Clusters and Databricks Clusters