Metadata-Version: 2.1
Name: tactis
Version: 0.1.1
Summary: Transformer-Attentional Copulas for Multivariate Time Series
Home-page: https://github.com/ServiceNow/tactis/tree/tactis-2
Author: Arjun Ashok, Étienne Marcotte, Valentina Zantedeschi, Nicolas Chapados, Alexandre Drouin
Description-Content-Type: text/markdown
Provides-Extra: research
License-File: LICENSE
License-File: NOTICE

# TACTiS-2: Better, Faster, Simpler Attentional Copulas for Multivariate Time Series

Arjun Ashok, Étienne Marcotte, Valentina Zantedeschi, Nicolas Chapados, Alexandre Drouin (2023). *[TACTiS-2: Better, Faster, Simpler Attentional Copulas for Multivariate Time Series](https://arxiv.org/abs/2310.01327)*. (Preprint)

> We introduce a new model for multivariate probabilistic time series prediction, designed to flexibly address a range of tasks including forecasting, interpolation, and their combinations. Building on copula theory, we propose a simplified objective for the recently-introduced transformer-based attentional copulas (TACTiS), wherein the number of distributional parameters now scales linearly with the number of variables instead of factorially. The new objective requires the introduction of a training curriculum, which goes hand-in-hand with necessary changes to the original architecture. We show that the resulting model has significantly better training dynamics and achieves state-of-the-art performance across diverse real-world forecasting tasks, while maintaining the flexibility of prior work, such as seamless handling of unaligned and unevenly-sampled time series.

[[Preprint]](https://arxiv.org/abs/2310.01327)

<br />

<p align="center">
  <img src="https://github.com/ServiceNow/tactis/blob/tactis-2/cover.png?raw=true" width="500" />
</p>


## Installation

You can install the TACTiS-2 model with [pip](https://pip.pypa.io/):

```console
pip install tactis
```

Alternatively, the `research` version installs `gluonts` and `pytorchts` as dependencies which are required to replicate experiments from the paper:

```console
pip install tactis[research]
```

Note: `tactis` has been currently tested with Python 3.10.8.

## Instructions

With the `research` version of the code, [`train.py`](https://github.com/ServiceNow/tactis/blob/tactis-2/train.py) can be used to train the TACTiS-2 model for a specific dataset. The arguments in [`train.py`](https://github.com/ServiceNow/tactis/blob/tactis-2/train.py) can be used to specify the dataset, the training task (forecasting or interpolation), the hyperparameters of the model and a whole range of other training options.

There are notebooks in the that are useful in guiding training and evaluation pipeline setups: [`random_walk.ipynb`](https://github.com/ServiceNow/tactis/blob/tactis-2/demo/random_walk.ipynb) demonstrates TACTiS-2 on a simple low-dimensional random walk dataset, and [`gluon_fred_md_forecasting.ipynb`](https://github.com/ServiceNow/tactis/blob/tactis-2/demo/gluon_fred_md_forecasting.ipynb) demonstrates how to train and evaluate TACTiS-2 on the [FRED-MD dataset](https://zenodo.org/records/4654833) used in the paper. Note that the [`gluon_fred_md_forecasting.ipynb`](https://github.com/ServiceNow/tactis/blob/tactis-2/demo/gluon_fred_md_forecasting.ipynb) notebook requires GluonTS and PyTorchTS to be installed.


## Note

For an implementation of the [original version of TACTiS](https://arxiv.org/abs/2202.03528), please see [here](https://github.com/ServiceNow/tactis/tree/v1.0.0).

## Citing this work

Please use the following Bibtex entry to cite TACTiS-2.

```
@misc{ashok2023tactis2,
      title={TACTiS-2: Better, Faster, Simpler Attentional Copulas for Multivariate Time Series}, 
      author={Arjun Ashok and Étienne Marcotte and Valentina Zantedeschi and Nicolas Chapados and Alexandre Drouin},
      year={2023},
      eprint={2310.01327},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}
```
