Skip to content

Causality tests

The syntax for performing a causality tests is always the same: causality(source, target, test). Here, test is an instance of one of the concrete types listed below. Each of these are containers for test parameters. Returned values from causality depend on the test type - see the documentation for specific tests for details.

Uncertainty handling

All tests can be used in conjunction with a binning scheme with BinnedDataCausalityTest to perform the tests on data with uncertain index (e.g time) values. See also tutorials.

Distance-based causality tests

# CausalityTools.CausalityTests.DistanceBasedCausalityTest — Type.

DistanceBasedCausalityTest{N}

The supertype of all abstract and composite types representing a causality test based on some sort of distance computation.

The type parameter N indicates the number of returned elements when applying the test.

Concrete subtypes are

source

Entropy-based tests

# CausalityTools.CausalityTests.EntropyBasedCausalityTest — Type.

EntropyBasedCausalityTest

The supertype of all abstract and composite types representing a causality test based on some entropy based measure.

Concrete subtypes are those based on transfer entropy

and those that in some manner utilise transfer entropy or other information theoretic approaches :

source

Transfer entropy tests

# CausalityTools.CausalityTests.TransferEntropyCausalityTest — Type.

TransferEntropyCausalityTest{N}

The supertype of all abstract and composite types representing a transfer entropy causality test applied to N different prediction lags.

Concrete subtypes are

source