Darts: User-Friendly Modern Machine Learning for Time Series
Authors: Julien Herzen, Francesco Lässig, Samuele Giuliano Piazzetta, Thomas Neuer, Léo Tafti, Guillaume Raille, Tomas Van Pottelbergh, Marek Pasieka, Andrzej Skrodzki, Nicolas Huguenin, Maxime Dumonal, Jan Kościsz, Dennis Bader, Frédérick Gusset, Mounir Benheddi, Camila Williamson, Michal Kosinski, Matej Petrik, Gaël Grosch
JMLR 2022 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The code below shows how to fit a single TCN model (Bai et al., 2018) with default hyperparameters on two different (and completely distinct) series, and forecast one of them. The network outputs the parameters of a Laplace distribution. The code contains a complete predictive pipeline, from loading and preprocessing the data, to plotting the forecast with arbitrary quantiles (shown on the right). |
| Researcher Affiliation | Industry | Julien Herzen EMAIL Francesco L assig EMAIL Samuele Giuliano Piazzetta EMAIL Thomas Neuer EMAIL L eo Tafti EMAIL Guillaume Raille EMAIL Tomas Van Pottelbergh EMAIL Marek Pasieka EMAIL Andrzej Skrodzki EMAIL Nicolas Huguenin EMAIL Maxime Dumonal EMAIL Jan Ko scisz EMAIL Dennis Bader EMAIL Fr ed erick Gusset EMAIL Mounir Benheddi EMAIL Camila Williamson EMAIL Michal Kosinski EMAIL Matej Petrik EMAIL Ga el Grosch EMAIL Unit8 SA, Switzerland |
| Pseudocode | No | No explicit pseudocode or algorithm blocks are provided. Section 3 contains a Python code snippet as a usage example, which is not pseudocode. |
| Open Source Code | Yes | We present Darts1, a Python machine learning library for time series, with a focus on forecasting. ... 1. https://github.com/unit8co/darts |
| Open Datasets | Yes | from darts.datasets import Air Passengers Dataset, Monthly Milk Dataset air = Air Passengers Dataset().load() milk = Monthly Milk Dataset().load() |
| Dataset Splits | No | The paper demonstrates fitting a model to `[air_s, milk_s]` and then predicting `n=48` steps for `air_s`. However, it does not specify any training/validation/test splits, percentages, or methodology for splitting these datasets. |
| Hardware Specification | No | All neural networks are implemented using Py Torch (Paszke et al., 2019) and support training and inference on GPUs. |
| Software Dependencies | No | All neural networks are implemented using Py Torch (Paszke et al., 2019)... similar to scikit-learn (Pedregosa et al., 2011)... Time Series are wrapping around three-dimensional xarray Data Array (Hoyer and Hamman, 2017)... convert to/from other common types, such as Pandas Dataframes or Num Py arrays (Harris et al., 2020). |
| Experiment Setup | Yes | model = TCNModel(input_chunk_length=24, output_chunk_length=12, likelihood=LL()) model.fit([air_s, milk_s], epochs=100) |