This repository provides the code for the paper Towards Better Forecasting by Fusing Near and Distant Future Visions, accepted by AAAI 2020.

Please install Git Large File Storage first and then use

git lfs clone https://github.com/smallGum/MLCNN-Multivariate-Time-Series.git

to download all the datasets and codes.

You can find the dataset in the data/ folder.

Examples with parameter grid search to run different datasets are in runTraffic.py, runEnergy.py and runNASDAQ.py.

Python 3.6.7 and Pytorch 1.0.0

This multivariate time series forecasting framework was implemented based on the following two repositories:

• LSTNet
• SOCNN

Some codes and design patterns are borrowed from these two excellent frameworks.