plot.btergmgof: Plot or print btergmgof objects

Description

Plot or print formatted goodness-of-fit statistics and degeneracy checks from btergmgof objects.

Usage

## S3 method for class 'btergmgof':
plot(x, boxplot = TRUE, boxplot.mfrow = TRUE, 
    boxplot.dsp = TRUE, boxplot.esp = TRUE, boxplot.geodist = TRUE, 
    boxplot.degree = TRUE, boxplot.idegree = TRUE, 
    boxplot.odegree = TRUE, boxplot.kstar = TRUE, boxplot.istar = TRUE, 
    boxplot.ostar = TRUE, boxplot.dsp.max = NULL, 
    boxplot.esp.max = NULL, boxplot.geodist.max = NULL, 
    boxplot.degree.max = NULL, boxplot.idegree.max = NULL, 
    boxplot.odegree.max = NULL, boxplot.kstar.max = NULL, 
    boxplot.istar.max = NULL, boxplot.ostar.max = NULL, 
    boxplot.transform = function(x) x, boxplot.border = "darkgray", 
    boxplot.mean.col = "black", boxplot.median.col = "black", 
    boxplot.lwd = 0.8, boxplot.outline = FALSE, boxplot.ylab = "Frequency", 
    boxplot.main = NULL, boxplot.ylim = NULL, roc = TRUE, pr = TRUE, 
    rocpr.add = FALSE, rocpr.avg = c("none", "horizontal", "vertical", 
    "threshold"), rocpr.spread = c("boxplot", "stderror", "stddev"), 
    rocpr.lwd = 3, roc.main = NULL, roc.random = FALSE, roc.col = "#bd0017", 
    roc.random.col = "#bd001744", pr.main = NULL, pr.random = FALSE, 
    pr.col = "#5886be", pr.random.col = "#5886be44", pr.poly = 0, ...)
## S3 method for class 'btergmgof':
print(x, classicgof = TRUE, 
    rocprgof = TRUE, degeneracy = TRUE, ...)
## S3 method for class 'btergmgof':
summary(object, classicgof = TRUE, 
    rocprgof = TRUE, degeneracy = TRUE, ...)

Arguments

A btergmgof object created by calling the gof.btergm method.

boxplot

Create a plot where simulated statistics are summarized as boxplots and observed networks as lines (= classic statnet-style GOF plots)?

boxplot.mfrow

Should the classic GOF plots come out separately (boxplot.mfrow = FALSE), or should all statistics be aligned in a single diagram (boxplot.mfrow = TRUE)? Returning the plots separately can be helpful if the output is redirected t

boxplot.dsp

Plot a GOF diagram for dyad-wise shared partners?

boxplot.esp

Plot a GOF diagram for edge-wise shared partners?

boxplot.geodist

Plot a GOF diagram for geodesic distances?

boxplot.degree

Plot a GOF diagram for the degree distribution?

boxplot.idegree

Plot a GOF diagram for the indegree distribution?

boxplot.odegree

Plot a GOF diagram for the outdegree distribution?

boxplot.kstar

Plot a GOF diagram for the kstar distribution?

boxplot.istar

Plot a GOF diagram for the instar distribution?

boxplot.ostar

Plot a GOF diagram for the outstar distribution?

boxplot.dsp.max

Upper bound of dyad-wise shared partners to report.

boxplot.esp.max

Upper bound of edge-wise shared partners to report.

boxplot.geodist.max

Upper bound of geodesic distances to report (excluding Inf values).

boxplot.degree.max

Upper bound of the degree statistic to report.

boxplot.idegree.max

Upper bound of the indegree statistic to report.

boxplot.odegree.max

Upper bound of the outdegree statistic to report.

boxplot.kstar.max

Upper bound of the kstar statistic to report.

boxplot.istar.max

Upper bound of the istar statistic to report.

boxplot.ostar.max

Upper bound of the ostar statistic to report.

boxplot.transform

A function which transforms the y values used for the boxplots. For example, if some of the values become very large and make the output illegible, boxplot.transform = function(x) x^0.1 or a similar transformation of the values can be used. N

boxplot.border

Color of the borders of the boxplots.

boxplot.mean.col

Color of the mean of the observed network statistic.

boxplot.median.col

Color of the median of the observed network statistic.

boxplot.lwd

Scaling factor for the line width of the boxplots and of the line for the observed network statistics.

boxplot.outline

Print outliers in the boxplots?

boxplot.ylab

Label of the y-axis of a GOF plot.

boxplot.main

Main title of a GOF plot. Note: The same main title is used for all resulting diagrams. To obtain diagrams with varying main titles, plot each statistic separately using a new call of the plot.btergm method.

boxplot.ylim

Vertical limit of the boxplots. Only the maximum value must be provided because the minimum is always 0. Providing a vertical limit may be sensible if multiple models have to be compared side by side.

roc

Should the ROC curve(s) be plotted?

Should the precision-recall curve(s) be plotted?

rocpr.add

Add the ROC and/or PR curve to an existing plot?

rocpr.avg

Averaging pattern for the ROC and PR curve(s) if multiple target time steps were used. Allowed values are "none" (plot all curves separately), "horizontal" (horizontal averaging), "vertical" (vertical averaging), and

rocpr.spread

When multiple target time steps are used and curve averaging is enabled, the variation around the average curve can be visualized as standard error bars ("stderror"), standard deviation bars ("stddev"), or by using box plots (

rocpr.lwd

Line width of the ROC and PR curve(s).

roc.main

Main title of the ROC plot.

roc.random

Draw an additional ROC curve for random graphs with the same tie probability (as a reference value)?

roc.col

Color of the ROC curve(s). By default, a dark red color is used.

roc.random.col

Color of the random graph ROC curve.

pr.main

Main title of the ROC plot.

pr.random

Draw an additional PR curve for random graphs with the same tie probability (as a reference value)?

pr.col

Color of the precision-recall curve(s). By default, a blue color is used.

pr.random.col

Color of the random graph PR curve.

pr.poly

If a value of 0 is set, nothing special happens. If a value of 1 is set, a straight line is fitted through the PR curve and displayed. Values between 2 and 9 fit higher-order polynomial curves through th

classicgof

Print tables with t-tests comparing observed and simulated networks, as in statnet?

rocprgof

Print tables with the area under the curve (AUC) for the PR and ROC curves at each time step of the observed network(s)?

degeneracy

Print tables with t-tests comparing the global statistics of observed and simulated networks for each observed time step? This is essentially a degeneracy check.

object

A btergmgof object created by calling the gof.btergm method.

...

Further arguments. These arguments are passed on to the plot command. This facilitates customization of the plots.

Details

These methods plot or print nicely formatted goodness-of-fit statistics from btergmgof objects to the R console. The typical workflow is to estimate a TERGM using the btergm function and save it as an object, then use the gof function to produce a btergmgof object, and finally use the functions and methods described here to show the output of this resulting object to assess the goodness of fit.

The plot method plots (1) boxplots of network statistics and draws the observed statistics as lines (classic statnet-like GOF boxplots) and (2) receiver operating characteristics (ROC) and precision recall (PR) curves.

The print and summary methods show (1) observed versus simulated network statistics (classic statnet-like GOF tables), (2) the area under the ROC curve (AUC) for each observed network, and (3) degeneracy checks by comparing global statistics of simulated versus observed networks.

Description

Usage

Arguments

Details

See Also