EGA {EGAnet} R Documentation

Applies the Exploratory Graph Analysis technique

Description

Estimates the number of dimensions of a given dataset or correlation matrix using the graphical lasso (EBICglasso.qgraph) or the Triangulated Maximally Filtered Graph (TMFG) network estimation methods.

Usage

EGA(
data,
n = NULL,
uni.method = c("expand", "LE"),
corr = c("cor_auto", "pearson", "spearman"),
model = c("glasso", "TMFG"),
model.args = list(),
algorithm = c("walktrap", "louvain"),
algorithm.args = list(),
plot.EGA = TRUE,
plot.type = c("GGally", "qgraph"),
plot.args = list(),
verbose = TRUE,
...
)


Arguments

 data Matrix or data frame. Variables (down columns) or correlation matrix. If the input is a correlation matrix, then argument n (number of cases) is required n Integer. Sample size if data provided is a correlation matrix uni.method Character. What unidimensionality method should be used? Defaults to "LE". Current options are: expand Expands the correlation matrix with four variables correlated .50. If number of dimension returns 2 or less in check, then the data are unidimensional; otherwise, regular EGA with no matrix expansion is used. This is the method used in the Golino et al. (2020) Psychological Methods simulation. LE Applies the leading eigenvalue algorithm (cluster_leading_eigen) on the empirical correlation matrix. If the number of dimensions is 1, then the leading eigenvalue solution is used; otherwise, regular EGA is used. This is the final method used in the Christensen, Garrido, and Golino (2021) simulation. corr Type of correlation matrix to compute. The default uses cor_auto. Current options are: cor_auto Computes the correlation matrix using the cor_auto function from qgraph. pearson Computes Pearson's correlation coefficient using the pairwise complete observations via the cor function. spearman Computes Spearman's correlation coefficient using the pairwise complete observations via the cor function. model Character. A string indicating the method to use. Defaults to "glasso". Current options are: glasso Estimates the Gaussian graphical model using graphical LASSO with extended Bayesian information criterion to select optimal regularization parameter TMFG Estimates a Triangulated Maximally Filtered Graph model.args List. A list of additional arguments for EBICglasso.qgraph or TMFG algorithm A string indicating the algorithm to use or a function from igraph Defaults to "walktrap". Current options are: walktrap Computes the Walktrap algorithm using cluster_walktrap louvain Computes the Louvain algorithm using cluster_louvain algorithm.args List. A list of additional arguments for cluster_walktrap, cluster_louvain, or some other community detection algorithm function (see examples) plot.EGA Boolean. If TRUE, returns a plot of the network and its estimated dimensions. Defaults to TRUE plot.type Character. Plot system to use. Current options are qgraph and GGally. Defaults to "GGally" plot.args List. A list of additional arguments for the network plot. For plot.type = "qgraph": vsize Size of the nodes. Defaults to 6. For plot.type = "GGally" (see ggnet2 for full list of arguments): vsize Size of the nodes. Defaults to 6. label.size Size of the labels. Defaults to 5. alpha The level of transparency of the nodes, which might be a single value or a vector of values. Defaults to 0.7. edge.alpha The level of transparency of the edges, which might be a single value or a vector of values. Defaults to 0.4. legend.names A vector with names for each dimension color.palette The color palette for the nodes. For custom colors, enter HEX codes for each dimension in a vector. See color_palette_EGA for more details and examples verbose Boolean. Should network estimation parameters be printed? Defaults to TRUE. Set to FALSE for no print out ... Additional arguments. Used for deprecated arguments from previous versions of EGA

Details

Two community detection algorithms, Walktrap (Pons & Latapy, 2006) and Louvain (Blondel et al., 2008), are pre-programmed because of their superior performance in simulation studies on psychological data generated from factor models (Christensen & Golino; 2020; Golino et al., 2020). Notably, any community detection algorithm from the igraph can be used to estimate the number of communities (see examples).

Value

Returns a list containing:

 network A symmetric network estimated using either the EBICglasso.qgraph or TMFG wc A vector representing the community (dimension) membership of each node in the network. NA values mean that the node was disconnected from the network n.dim A scalar of how many total dimensions were identified in the network cor.data The zero-order correlation matrix

Author(s)

Hudson Golino <hfg9s at virginia.edu>, Alexander P. Christensen <alexpaulchristensen at gmail.com>, Maria Dolores Nieto <acinodam at gmail.com> and Luis E. Garrido <garrido.luiseduardo at gmail.com>

References

# Louvain algorithm
Blondel, V. D., Guillaume, J.-L., Lambiotte, R., & Lefebvre, E. (2008). Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment, 2008, P10008.

# Compared all igraph community detections algorithms, introduced Louvain algorithm, simulation with continuous and polytomous data
# Also implements the Leading Eigenvalue unidimensional method
Christensen, A. P., Garrido, L. E., & Golino, H. (2021). Comparing community detection algorithms in psychological data: A Monte Carlo simulation. PsyArXiv.

# Original simulation and implementation of EGA
Golino, H. F., & Epskamp, S. (2017). Exploratory graph analysis: A new approach for estimating the number of dimensions in psychological research. PLoS ONE, 12, e0174035.

Golino, H. F., & Demetriou, A. (2017). Estimating the dimensionality of intelligence like data using Exploratory Graph Analysis. Intelligence, 62, 54-70.

# Current implementation of EGA, introduced unidimensional checks, continuous and dichotomous data
Golino, H., Shi, D., Christensen, A. P., Garrido, L. E., Nieto, M. D., Sadana, R., & Thiyagarajan, J. A. (2020). Investigating the performance of Exploratory Graph Analysis and traditional techniques to identify the number of latent factors: A simulation and tutorial. Psychological Methods, 25, 292-320.

# Walktrap algorithm
Pons, P., & Latapy, M. (2006). Computing communities in large networks using random walks. Journal of Graph Algorithms and Applications, 10, 191-218.

bootEGA to investigate the stability of EGA's estimation via bootstrap and CFA to verify the fit of the structure suggested by EGA using confirmatory factor analysis.

Examples

# Estimate EGA
## plot.type = "qqraph" used for CRAN checks
## plot.type = "GGally" is the default
ega.wmt <- EGA(data = wmt2[,7:24], plot.type = "qgraph")

# Summary statistics
summary(ega.wmt)

# Estimate EGAtmfg
ega.wmt <- EGA(data = wmt2[,7:24], model = "TMFG", plot.type = "qgraph")

# Estimate EGA with Louvain algorithm
ega.wmt <- EGA(data = wmt2[,7:24], algorithm = "louvain", plot.type = "qgraph")

# Estimate EGA with Spinglass algorithm
ega.wmt <- EGA(data = wmt2[,7:24],
algorithm = igraph::cluster_spinglass, plot.type = "qgraph")

# Estimate EGA
ega.intel <- EGA(data = intelligenceBattery[,8:66], model = "glasso", plot.EGA = FALSE)

# Summary statistics
summary(ega.intel)



[Package EGAnet version 1.1.0 Index]