Multivariate Transfer Entropy estimation - pwollstadt/IDTxl GitHub Wiki

Import Multivariate TE class

from idtxl.multivariate_te import MultivariateTE

Import or generate data

from idtxl.data import Data

IDTxl uses its own class to handle data. Read the dedicated tutorial on The Data Object

Set up analysis options dictionary

The following sections will guide you through the set up of the analysis options dictionary.

Set CMI estimator

For mTE estimation the only required option is the name of the conditional mutual information (CMI) estimator. To learn how to set a specific estimator and for a list of the available estimators, read the dedicated tutorial on Estimators.

Set parameters for statistical tests

The parameters for the statistical significance tests are set to default values, unless specified by the user. Read the dedicated tutorial on Statistical Significance Tests.

Conditioning on specific variables

Read the dedicated tutorial on Conditioning on specific variables.

Initialise analysis object

Once the options dictionary is defined, create an analysis object for mTE estimation.

Set source and target lags

The MultivariateTE class requires the setting of the maximum source lag j, minimum source lag k, in addition to the analysis options defined before. To learn how to set the lags, read the dedicated tutorial on Source and Target Lags.

Run analysis

It is both possible to run a Single Target mTE analysis or a Network mTE analysis. Read the tutorials to learn how to set the source and target nodes.

Results

MultivariateTE returns a dictionary with results for the mTE estimation. In general, there is a results dictionary for each target. If multiple targets were analysed by calling analyse_network(), the output is a dictionary of dictionaries with one entry per analysed target. Read the tutorial on Results Dictionary for examples and an explanation of the dictionary entries.