ICausalDiscoveryAlgorithm<T>

Interfaces · AiDotNet.CausalDiscovery

Interface for causal structure learning algorithms that discover Directed Acyclic Graphs (DAGs) from data.

Known implementations: CausalDiscoveryBase<T>

For Beginners

This interface defines the contract for algorithms that figure out cause-and-effect relationships from data. Given a dataset with multiple variables, the algorithm produces a graph showing which variables directly cause changes in other variables.

For example, given data about weather, traffic, and commute time, a causal discovery algorithm might find: Weather → Traffic → Commute Time (weather causes traffic, which causes longer commutes).

How It Works

Causal discovery algorithms analyze observational data to infer the causal structure — a DAG where edges represent direct causal relationships between variables. Unlike correlation analysis, these algorithms attempt to determine the direction of causation.

Properties

Property Summary
CausalDiscoveryCategory Category { get; } Gets the methodological category of this algorithm.
bool SupportsLatentConfounders { get; } Gets whether this algorithm can handle latent (unobserved) confounders.
bool SupportsMixedData { get; } Gets whether this algorithm supports mixed (continuous and discrete) data types.
bool SupportsNonlinear { get; } Gets whether this algorithm can discover nonlinear causal relationships.
bool SupportsTimeSeries { get; } Gets whether this algorithm is designed for time series data.
string Name { get; } Gets the display name of this algorithm.

Methods

Method Summary
CausalGraph<T> DiscoverStructure(Matrix<T> data, Vector<T> target, string[] featureNames) Discovers causal structure with a designated target variable for directed analysis.
CausalGraph<T> DiscoverStructure(Matrix<T> data, string[] featureNames) Discovers causal structure from an observational data matrix.