

Random chunk resampling

Note

Uncertain data handling relies on the UncertainData. To use the resampling schemes, you need to load this package by first running using UncertainData in the Julia console.

RandomSequencesTest

# CausalityTools.CausalityTests.RandomSequencesTest — Type.

RandomSequencesTest(test::CausalityTest, sequences_resampling::RandomSequences)

A causality test where the test is applied to multiple independent draws of the datasets, where each draw from a randomly selected consecutive chunk of points.

Examples

Say we want do do a cross mapping test on a set of time series with N = 200 points. To get some ensemble statistics, we want to do an ensemble analysis on 150 randomly selected chunks of the time series with lengths ranging from N-30 to N-1.

# A cross mapping test with default parameters (don't use default parameters in a real analysis!)
cm_test = CrossMappingTest()

# Combine with the random sequence subsamling
rtest = RandomSequencesTest(cm_test, RandomSequences(150, N-30:N-1))

source

Call signature

# CausalityToolsBase.causality — Method.

causality(source, target, test::RandomSequencesTest)

Apply a causality test on random, consecutive chunks of x and y.

The chunk length can be fixed (an integer) or a collection of chunk lengths.

Works on any inputs x and y, both scalar vectors and uncertain datasets.

Example

For scalar valued vectors:

using CausalityTools, UncertainData

x, y = rand(100), rand(100)

# Use a cross-mapping test with default parameters (not recommended,
# make sure you understand the parameters!)
cm_test = CrossMappingTest()

# A random sequences test that applies the cross mapping test 
# to 150 different random chunks of the data of lengths 80 to 90
n_realisations, chunk_lengths = 150, 80:90
rs = RandomSequences(n_realisations, chunk_lengths)

# Combine to a RandomSequencesTest
rs_test = RandomSequencesTest(cm_test, chunk_lengths)

# Compute causality statistic
causality(x, y, rs_test)

This also works on uncertain data, or any combination of scalar vectors and uncertain data:

using CausalityTools, UncertainData

# Some example data
N = 300
sys = ar1_unidir(c_xy = 0.8)
X, Y = example_uncertain_indexvalue_datasets(sys, N, (1, 2),
    d_xval = Uniform(0.001, 0.05), d_yval = Uniform(0.001, 0.05));

# Apply cross mapping test on 100 different randomly selected chunks
# of lengths 75 to 90
r = RandomSequences(100, 75:90)
rs_test = RandomSequencesTest(CrossMappingTest(), r)
causality(X, Y, rs_test)

source