This repo is a Pytorch implementation of the paper--RAND:Adaptive Normalization and Denormalization Method for Non-Stationary Time Series Forecasting.
RAND is a plug-and-play normalization and denormalization method, namely Resolution-Adaptive Normalization and Denormalization, it is devised to deal with the distribution shift problem in time series for machine-learning-based forecasting model. It normalizes the input time series to reduce distribution differences between instances and addaptively denormalizes the output series by modeling variations of slice-level time-varying mean and variance.
Ensure you are using Python 3.9 and install the necessary dependencies by running:
pip install -r requirements.txt
mkdir datasetsAll the datasets are available at the Google Driver provided by Autoformer. Begin by downloading the required datasets. All datasets are conveniently available at Autoformer. Create a separate folder named ./dataset and neatly organize all the csv files as shown below:
dataset
└── electricity.csv
└── ETTh1.csv
└── ETTh2.csv
└── ETTm1.csv
└── ETTm2.csv
└── traffic.csv
└── weather.csv
We provide ready-to-use scripts for RAND enhanced backbone models.
sh run_rand.shSpecial thanks to the following repositories for their invaluable code and datasets:
https://github.com/thuml/Autoformer
https://github.com/honeywell21/DLinear
https://github.com/cure-lab/LTSF-Linear
https://github.com/icantnamemyself/SAN
https://github.com/wanghq21/MICN
https://github.com/thuml/Time-Series-Library
https://github.com/MAZiqing/FEDformer
https://github.com/zhouhaoyi/Informer2020
https://github.com/weifantt/Dish-TS
https://github.com/yuqinie98/PatchTST
https://github.com/Thinklab-SJTU/Crossformer
If you have any questions, please contact xiel21@mails.tsinghua.edu.cn or submit an issue.