Package 'circularNet'

Title: Circular Graphical Model Estimation
Description: Provides methods for estimating circular graphical models using maximum likelihood estimation (MLE) and circular mean squared error (CMSE) approaches. The package includes tools for model fitting, network construction, network evaluation, and visualization. The CMSE-based methodology is related to Dar (2023) <https://open.metu.edu.tr/handle/11511/102577>. The package supports both simulated circular data and real-world applications, including gene-expression network analysis.
Authors: Fatemeh Sarasir [aut, cre] (ORCID: <https://orcid.org/0009-0003-5432-1352>), Vilda Purutcuoglu [aut] (ORCID: <https://orcid.org/0000-0002-3913-9005>), Elif Dogan Dar [aut] (ORCID: <https://orcid.org/0000-0001-7969-3767>)
Maintainer: Fatemeh Sarasir <[email protected]>
License: MIT + file LICENSE
Version: 0.1.0
Built: 2026-07-15 09:48:02 UTC
Source: https://github.com/cran/circularNet

Help Index

Build Adjacency Matrix

Description

Constructs a binary network from estimated coefficients using thresholding.

Usage

build_network(beta_hat, threshold = 0.5)

Arguments

beta_hat

Coefficient matrix.
threshold

Threshold value.

Value

Binary adjacency matrix.

Examples

set.seed(1)

data <- matrix(
runif(100, -pi, pi),
ncol = 5
)

fit <- fit_circular_model(data)

build_network(
fit,
threshold = 0.2
)

Circular Mean Squared Error

Description

Computes the circular mean squared error.

Usage

CMSE(par, data)

Arguments

par

Parameter vector.
data

Data matrix (p x n).

Value

Numeric CMSE value.

Examples

data <- matrix(
runif(20, -pi, pi),
nrow = 4
)

par <- rep(0, nrow(data))

CMSE(par, data)

Evaluate Estimated Network

Description

Computes performance metrics comparing an estimated network with a reference network.

Usage

evaluate_network(est, truth)

Arguments

est

Estimated adjacency matrix.
truth

Ground-truth adjacency matrix.

Value

List of evaluation metrics.

Examples

est <- matrix(
c(0,1,0,
1,0,1,
0,1,0),
nrow = 3,
byrow = TRUE
)

truth <- matrix(
c(0,1,0,
1,0,0,
0,0,0),
nrow = 3,
byrow = TRUE
)

evaluate_network(est, truth)

Fit Circular Graphical Model

Description

Estimates a circular graphical model using node-wise regression and likelihood-based optimization.

Usage

fit_circular_model(data)

Arguments

data

Matrix of circular observations (rows = observations, columns = variables).

Value

A coefficient matrix.

Examples

set.seed(1)

data <- matrix(
runif(100, -pi, pi),
ncol = 5
)

fit <- fit_circular_model(data)
round(fit, 2)

Log-Likelihood for Circular Model

Description

Computes the negative log-likelihood used for parameter estimation in the likelihood-based circular graphical model.

Usage

log_likelihood(params, theta, phi, mu_theta, mu_phi)

Arguments

params

Parameter vector containing alpha, beta coefficients, and kappa.
theta

Response variable.
phi

Predictor matrix.
mu_theta

Mean of theta.
mu_phi

Mean of phi.

Value

Negative log-likelihood value.

Examples

theta <- runif(20, -pi, pi)
phi <- matrix(runif(40, -pi, pi), ncol = 2)
params <- c(0.1, 0.1, 0.1, 1)

log_likelihood(
params,
theta,
phi,
mean(theta),
colMeans(phi)
)

Wrap Angles to (-pi, pi)

Description

Transforms numeric values to the interval (-pi, pi).

Usage

mod_twopi(x)

Arguments

x

Numeric vector of angles.

Value

Numeric vector of wrapped angles.

Examples

mod_twopi(c(-4 * pi, -3, 0, 3, 4 * pi))

Fit Circular Model Using CMSE

Description

Fits a circular regression model using CMSE optimization.

Usage

model_fit_cmse(data)

Arguments

data

Data matrix (p x n), where rows correspond to variables and columns correspond to samples.

Value

A list containing:

  • beta: coefficient matrix

  • code: optimization convergence codes

Examples

set.seed(1)

data <- matrix(
  runif(40, -pi, pi),
  nrow = 4
)

model_fit_cmse(data)

Plot Estimated Network

Description

Visualizes an adjacency matrix as a network graph.

Usage

plot_network(adj_matrix)

Arguments

adj_matrix

Adjacency matrix.

Value

A network plot.

Examples

adj <- matrix(
c(0,1,1,
1,0,0,
1,0,0),
nrow = 3,
byrow = TRUE
)

plot_network(adj)

Compute Residuals

Description

Computes predicted and observed circular values.

Usage

residual_fnc(par, data)

Arguments

par

Parameter vector.
data

Data matrix (p x n).

Value

List with predicted and observed values.

Examples

data <- matrix(
runif(20, -pi, pi),
nrow = 4
)

par <- rep(0, nrow(data))

residual_fnc(par, data)