Kristian Agasøster Haaga

WIP: Compression complexity causality

6

## What is this PR? This PR implements the effort-to-compress algorithm that underlies the Compression-Complexity Causality (CCC) in #161. It also implements the CCC algorithm in #161. ## TODO -...

Penchants and leanings

3

An implementation of the penchants and leanings algorithms from McCracken and Weigel (2016), which is meant for exploratory causal inference. Addresses #28. The docs are [here](https://juliadynamics.github.io/CausalityTools.jl/previews/PR153).

Simplex projection

4

An implementation of the simplex projection algorithm from Sugihara and May (1990), as part of some more cross mapping stuff that is coming. The docs are [here](https://juliadynamics.github.io/CausalityTools.jl/previews/PR150/edm/). This probably belongs...

Add transfer entropy example with modern estimators

1

The entropy-based estimators have larger biases, the modern estimators such as `Lindner` and `Zhu1` have less bias. We should include an example in the docs showcasing why they should be...

A common use case for `independence` is to use `SurrogateTest` and `LocalPermutationTest` to test for independence between two or more variables using some information theoretic (conditional) independence measure. Currently, for...

Clarify the test/measure/estimator setup better for the nonparametric case

Note: Combining this with an internal change to the type system, where we differ between `DirectionalAssociationMeasure` and `AssociationMeasure` (implicitly non-directional) massively simplifies code for the upcoming PC-algorithms.

TimeseriesSurrogates.jl, from version 2.2 onwards, implements `SurrogateTest`. We need to rename our `SurrogateTest` and provide a deprecation warning that points users towards the new name.

Preferred direction of flow

2

Implement a measure for the preferred all bivariate directional association measures. - [ ] Transfer entropy (TE). This will be a wrapper function that computes normalised TE (NTE) in both...

Currently, `CrossmapMeasure` constructors only accept fixed embedding parameters (dimension and lag). There is no reason to restrict embeddings like this. As suggested by @norbertgenera in #177, we should introduce generic...