Thread by @MergenthalerMax

In today's exciting release😎

Use @nixtlainc's StatsForecast to beat a WIDE variety of #DeepLearning models with:

- interpretable methods 📈
- in under 10 min ⏲️
- $0.5c in AWS 🔥
- and a few lines of #python code 🤯

#reinventforecasting

🧵 1/5

🧵2/5
Building upon the great work of @spyrosmakrid et al, we fitted an ensemble of simple statistical models:

AutoARIMA, Exponential Smoothing (@robjhyndman etal),

Complex Exponential Smoothing (@iSvetunkov etal)

and the DOT method (@fotpetr etal)

The results were great!🔥

🧵3/5
The simple statistical ensemble:

- outperforms most #DeepLearning models 🔥

- is 25,000 faster ⚡️

- slightly less accurate than an #DeepLearning ensemble 🤔

🧵4/5
In other words...

Deep-learning ensembles outperform statistical ensembles just by 0.36 points in SMAPE. 🔎

However, the DL ensemble:
- takes 14 days to run ⌛️
- and costs around USD 11,000 💸

The statistical ensemble takes:
- 6 min to run ⚡️
- and costs $0.5c 🎉

🧵5/5
You can reproduce the full experiment here: github.com/Nixtla/statsfo…

Please give us a ⭐️ if you find our repo useful