Source code for the paper, "DEPTS: Deep Expansion Learning for Periodic Time Series Forecasting", in ICLR22 Spotlight.
DEPTS is a customized deep neural network architecture for periodic time series forecasting, which aims to solve the following two challenges:
- To capture diversified periodic compositions
- To model complicated periodic dependencies
You can download the five benchmarks from Google Drive. All the datasets are well pre-processed. More details of datasets can be found in the paper. After downloading the zip file, please unzip it to the root dir of DEPTS for experiments.
Please use Python 3(.6) as well as the following packages:
torch >= 1.6.0
dataclasses
dtaidistance
pandas
numpy
tqdm
To reproduce the results, you can see more details in command.sh and directly run:
sh command.sh
Note that all the results reported in the paper are ensembled results of 30 models in order to get a robust evaluation and compare with N-BEATS. You can also try to run the single model for evaluation if you find it challenging to run all the models.
To get the evaluation results, run
python evaluation.py
If you find our work interesting, you can cite the paper as
@inproceedings{
fan2022depts,
title={{DEPTS}: Deep Expansion Learning for Periodic Time Series Forecasting},
author={Wei Fan and Shun Zheng and Xiaohan Yi and Wei Cao and Yanjie Fu and Jiang Bian and Tie-Yan Liu},
booktitle={International Conference on Learning Representations},
year={2022},
url={https://openreview.net/forum?id=AJAR-JgNw__}
}