PCMCIPlusAlgorithm<T>

Models & Types · AiDotNet.CausalDiscovery.TimeSeries

PCMCI+ — extension of PCMCI that also discovers contemporaneous causal links.

Inheritance: CausalDiscoveryBase<T> → TimeSeriesCausalBase<T> → PCMCIPlusAlgorithm<T>

For Beginners

PCMCI only finds “yesterday’s X causes today’s Y” relationships. PCMCI+ also finds “today’s X causes today’s Y” relationships, which are important when variables influence each other faster than the measurement interval.

How It Works

PCMCI+ extends PCMCI to handle both lagged AND contemporaneous (same time-step) causal links by adding a skeleton discovery and orientation step for lag-0 effects. It applies the same Fisher z-test based conditional independence testing at lag 0 conditioned on the lagged parents discovered by PCMCI.

Reference: Runge (2020), “Discovering Contemporaneous and Lagged Causal Relations in Autocorrelated Nonlinear Time Series Datasets”, UAI.

Constructors

Constructor Summary
PCMCIPlusAlgorithm(CausalDiscoveryOptions options)

Properties

Property Summary
bool SupportsNonlinear { get; }
string Name { get; }