List view
In this track we are trying to rework our current documentation to be better.
Overdue by 1 year(s)•Due by September 1, 2023•4/7 issues closed- Overdue by 2 year(s)•Due by December 31, 2022•1/3 issues closed
DL models backlog to be implemented: | name | TLDR | priority | | ------- | -------| --------| | [Autoformer](https://arxiv.org/pdf/2106.13008.pdf) | current best $\mathcal{O}( N \log N)$ transformer. Long horizon forecasts | | [N-Beats](https://arxiv.org/abs/1905.10437) | Uncommon net architecture. Beats es-rnn(M4 winner) on M4 | 🔥 | | [N-Hits](https://arxiv.org/pdf/2201.12886.pdf) | N-Beats predecessor. Beats Autoformer. Long horizon forecasts | 🔥 🔥 | [TFT](https://arxiv.org/pdf/1912.09363.pdf) | Interpretable time series forecasting. Transfromer bases. Probabilistic forecasts | 🥶 | | [DeepAR](https://arxiv.org/pdf/1704.04110.pdf) | RNN based. Probabilistic forecasts | 🥶 | | [NLinear](https://github.com/cure-lab/LTSF-Linear) | Beats Autoformer. Simple | 🔥 | | Vanilla Transformer | | 🔥 | | [DeepTime](https://github.com/salesforce/DeepTime)| | 🔥 | | [TIDE](https://ai.googleblog.com/2023/04/recent-advances-in-deep-long-horizon.html?m=1) | | 🔥 | | [TimesNet](https://openreview.net/pdf?id=ju_Uqw384Oq) | | 🔥 🔥 🔥 | [Metrics](https://docs.google.com/spreadsheets/d/1lELe6kbfzOI8vYOeVJC702b0JtCsz7OG9bvpnIk44JM/edit?usp=sharing) (There are few intersections or authors reused the same metrics from other papers without reproduced results)
No due date•6/7 issues closedRemove Transforms from TSDataset
No due date•30/31 issues closed