graph.param.estimator: Graph parameter estimator
In statGraph: Statistical Methods for Graphs

Description Usage Arguments Value References Examples

graph.param.estimator estimates the parameter that best approximates the model to the observed graph according to the Graph Information Criterion (GIC).

graph.param.estimator(
  G,
  model,
  parameters = NULL,
  eps = 0.01,
  bandwidth = "Silverman",
  eigenvalues = NULL,
  spectra = NULL,
  classic = FALSE
)

`G`	the undirected graph (igraph type) or its adjacency matrix. The adjacency matrix of an unweighted graph contains only 0s and 1s, while the weighted graph may have nonnegative real values that correspond to the weights of the edges.
`model`	either a string or a function: A string that indicates one of the following models: "ER" (Erdos-Renyi random graph), "GRG" (geometric random graph), "KR" (k regular random graph), "WS" (Watts-Strogatz model), and "BA" (Barabasi-Albert model). A function that returns a graph (represented by its adjacency matrix) generated by a graph model. It must contain two arguments: the first one corresponds to the graph size and the second to the parameter of the model.
`parameters`	numeric vector containing the values that that will be considered for the parameter estimation. The 'graph.param.estimator' will return the element of 'parameter' that minimizes the Kullback-Leiber divergence. If the user does not provide the argument 'parameters', and 'model' is an array, then the values considered for the parameter estimation are the rownames converted to a numeric vector. If 'model' is a string, then default values are used for the predefined models ("ER", "GRG", "KR", "WS", and "BA"). The default 'parameter' argument corresponds to a sequence from 0 to 1 with step 'eps' for the "ER" model (Erdos-Renyi random graph), in which the parameter corresponds to the probability to connect a pair of vertices; 0 to sqrt(2) with step 'eps' for the "GRG" model (geometric random graph), in which the parameter corresponds to the radius used to construct the geometric graph in a unit square; 0 to 'n' with step 'n*eps' for the "KR" model (k regular random graph), in which the parameter of the model corresponds to the degree 'k' of a regular graph; 0 to 1 with step 'eps' for the "WS" model (Watts-Strogatz model), in which the parameter corresponds to the probability to reconnect a vertex; and 0 to 3 with step 'eps' for the "BA" model (Barabasi-Albert model), in which the parameter corresponds to the scaling exponent.
`eps`	precision of the grid (default is 0.01) when 'classic' is TRUE.
`bandwidth`	string showing which criterion is used to choose the bandwidth during the spectral density estimation. Choose between the following criteria: "Silverman" (default), "Sturges", "bcv", "ucv" and "SJ". "bcv" is an abbreviation of biased cross-validation, while "ucv" means unbiased cross-validation. "SJ" implements the methods of Sheather & Jones (1991) to select the bandwidth using pilot estimation of derivatives.
`eigenvalues`	optional parameter. It contains the eigenvalues of matrix G. Then, it can be used when the eigenvalues of G were previously computed. If no value is passed, then the eigenvalues of G will be computed by 'graph.param.estimator'.
`spectra`	optional parameter containing the precomputed spectrum of the model. It is a three-dimensional array in which the first dimension corresponds to all parameters that will be explored in the grid, the second dimension has the same size of the given graph, and the third one corresponds to graphs randomly generated by the model. Thus, the position (i,j,k) contains the j-th eigenvalue of the k-th graph generated with the i-th parameter. The attribute 'rownames' of the array corresponds to the parameters converted to string. If spectra is NULL (default), then 'model' is used to generate random graphs and their spectra are computed automatically.
`classic`	logical. If FALSE (default) parameter is estimated using ternary search. If TRUE parameter is estimated using grid search.

A list containing:

`param`	the parameter estimate. For the "ER", "GRG", "KR", "WS", and "BA" models, the parameter corresponds to the probability to connect a pair of vertices, the radius used to construct the geometric graph in a unit square, the degree k of a regular graph, the probability to reconnect a vertex, and the scaling exponent, respectively.
`KLD`	the Kullback-Leibler divergence between the observed graph and the graph model with the estimated parameter.

Takahashi, D. Y., Sato, J. R., Ferreira, C. E. and Fujita A. (2012) Discriminating Different Classes of Biological Networks by Analyzing the Graph Spectra Distribution. _PLoS ONE_, *7*, e49949. doi:10.1371/journal.pone.0049949.

Silverman, B. W. (1986) _Density Estimation_. London: Chapman and Hall.

Sturges, H. A. The Choice of a Class Interval. _J. Am. Statist. Assoc._, *21*, 65-66.

Sheather, S. J. and Jones, M. C. (1991). A reliable data-based bandwidth selection method for kernel density estimation. _Journal of the Royal Statistical Society series B_, 53, 683-690. http://www.jstor.org/stable/2345597.

set.seed(1)
G <- igraph::sample_gnp(n=50, p=0.5)

# Using a string to indicate the graph model
result1 <- graph.param.estimator(G, "ER", eps=0.25)
result1

## Not run: 
# Using a function to describe the graph model
# Erdos-Renyi graph
set.seed(1)
model <- function(n, p) {
  return(igraph::sample_gnp(n, p))
}
result2 <- graph.param.estimator(G, model,  seq(0.2, 0.8, 0.1))
result2

## End(Not run)