This repository contains the source codes used in the following publications:
The script is mainly built using the following packages:
python 2.7.12 | Anaconda 2.3.0 (64-bit) (or higher) scikit-learn
In order to test the framework, a test file is provided (test.csv). This file contains all features used in the publication for one pixel on earth.
The script allows you to test both the "linear" and "non-linear" framework. In order to execute these, use the following commands:
python GC_script.py test.csv linear python GC_script.py test.csv non-linear
The outcome of both tests provide you information on the explained variance of the baseline and full model as well as a quantification of the Granger causality. For more information, see publication.
The output should look as follows:
Explained variance of baseline model: 0.076398 Explained variance of full model: 0.199819 Quantification of Granger causality: 0.123421 Total time: 58 seconds
Explained variance of baseline model: 0.055405 Explained variance of full model: 0.306413 Quantification of Granger causality: 0.251008 Total time: 6 seconds
This project is licensed under the GNU General Public License v3.0 - see the LICENSE file for details