Purpose

This project is the source code for the paper Time Series Forecasting based on High-Order Fuzzy Cognitive Maps and Wavelet Transform, which is now published in IEEE TFS.

Usage

• The file Wavelet_HFCM.py is the main program to perform forecasting time series by using Wavelet-HFCM.
• Defining the basic functions of an FCM, FCMs.py is used in the main program, and there is no need to run it seperately.
• The outcomes about the effects of two hyper-parameters (the order k and number of nodes Nc）on Wavelet-HFCM are saved into the file Outcome_for_papers/output_sunspot_sp500.xlsx, and their corresponding plots are saved into the directory ./Outcome_for_papers/impact_parameters/ .

Requirements

• python (3.6)
• matplotlib (3.0.3)
• seaborn (0.9.0)
• pandas (0.24.2)
• numpy (1.16.3)

Example

Here is an example for MG-chaos data.

Citation

If you find this work useful, please cite our paper:

@article{yang2018time,
  title={Time-series forecasting based on high-order fuzzy cognitive maps and wavelet transform},
  author={Yang, Shanchao and Liu, Jing},
  journal={IEEE Transactions on Fuzzy Systems},
  volume={26},
  number={6},
  pages={3391--3402},
  year={2018},
  publisher={IEEE}
}