estnet: Estimate the network parameters given an observed state...

In ghosoya/ludwig: Simulating and Estimating Undirected Binary Networks Using the Pseudolikelihood Method

Description

This function accepts a binary matrix of network states and estimates the thresholds and weights of the network model. The internally used functions are either cv.glmnet or glm. By default, cv.glmnet is used. In this case, the mixing parameter alpha is set to 0.5 (elastic net regularization) and lambda is chosen based on the solution with mimimum mean cross-validation error (argument s = "lambda.min" passed to function predict.cv.glmnet).

Usage

estnet(data, method = "glmnet", alpha = 0.5, s = "lambda.min", ...)

Arguments

data	a binary matrix of observed network states. The rows are the states, the columns are the nodes.
method	the method to be used. Either "glmnet" (default) or "glm".
alpha	the elastic mixing parameter passed to cv.glmnet.
s	value of the penalty parameter passed to predict.cv.glmnet.
...	other arguments passed over to e.g. cv.glmnet.

Value

The function returns a list of class estnet, containing...

• the objects returned by the function create_matrix.

• fitted fitted values based on the estimated parameters.

• y a vector of the observed node states.

• logL the log pseudo likelihood.

• deviance the deviance.

• coefficients the estimated coefficients (thresholds and weights).

• weights the estimated weights.

• thresholds the estimated thresholds.

• time the time used for the estimation process.

• net the list returned by either glm or cv.glmnet.

See Also

glmnet, cv.glmnet, predict.cv.glmnet, glm

ghosoya/ludwig documentation built on Sept. 20, 2020, 3:48 p.m.