pcMatrix: Pattern Causality Matrix Analysis

In patterncausality: Pattern Causality Algorithm

Pattern Causality Matrix Analysis

Description

Analyzes pattern causality relationships between multiple time series by computing pairwise causality measures and organizing them into matrices.

Usage

pcMatrix(
  dataset,
  E,
  tau,
  metric = "euclidean",
  h,
  weighted = TRUE,
  distance_fn = NULL,
  state_space_fn = NULL,
  relative = TRUE,
  verbose = FALSE,
  n_cores = 1
)

Arguments

dataset	Matrix or data frame of time series
E	Integer; embedding dimension
tau	Integer; time delay
metric	Character; distance metric ("euclidean", "manhattan", "maximum")
h	Integer; prediction horizon
weighted	Logical; whether to use weighted causality
distance_fn	Optional custom distance function
state_space_fn	Optional custom state space reconstruction function
relative	Logical; if TRUE calculates relative changes ((new-old)/old), if FALSE calculates absolute changes (new-old) in signature space. Default is TRUE.
verbose	Logical; whether to print progress
n_cores	Integer; number of cores for parallel computation

Details

Compute Pattern Causality Matrix Analysis

The function performs these key steps:

• Validates input data and parameters

• Computes pairwise causality measures

• Organizes results into causality matrices

• Provides summary statistics for each causality type

Value

A pc_matrix object containing causality matrices

Related Packages

• vars: Vector autoregression analysis

• tseries: Time series analysis tools

• forecast: Time series forecasting methods

patterncausality documentation built on April 3, 2025, 6:57 p.m.