Merge pull request #60 from snlab-nl/develop

Version 1.3.17

DESCRIPTION

@@ -2,8 +2,8 @@ Encoding: UTF-8
 Package: RSiena
 Type: Package
 Title: Siena - Simulation Investigation for Empirical Network Analysis
-Version: 1.3.16
-Date: 2023-01-02
+Version: 1.3.17
+Date: 2023-01-06
 Authors@R: c(person("Tom A.B.", "Snijders", role = c("cre", "aut"), email = "tom.snijders@nuffield.ox.ac.uk", comment = c(ORCID = "0000-0003-3157-4157")),
               person("Ruth", "Ripley", role = "aut"),
               person("Krists", "Boitmanis", role = c( "aut","ctb")),
@@ -38,7 +38,8 @@ Description: The main purpose of this package is to perform simulation-based
    There are also functions for testing parameters and checking goodness of fit.
    An overview of these models is given in Tom A.B. Snijders (2017), "Stochastic
    Actor-Oriented Models for Network Dynamics", Annual Review of Statistics and
-   Its Application, 4, 343-363 <doi:10.1146/annurev-statistics-060116-054035>.
+   Its Application, 4, 343-363 
+   <https://doi.org/10.1146/annurev-statistics-060116-054035>.
    An extensive manual, scripts, and much further information is at the Siena
    website <https://www.stats.ox.ac.uk/~snijders/siena/>.
 License: GPL-2 | GPL-3 | file LICENSE

NEWS.md

@@ -1,3 +1,17 @@
+# RSiena 1.3.17
+
+## 
+
+2023-01-06
+
+## Changes in RSiena:  
+
+### Improvements of functionality:
+  * `sienaGOF` now accepts simulated auxiliary statistics containing missing
+    values. If there are any, this will be reported with a warning
+    if `giveNAWarning` is `TRUE`.
+  * `sienaDataCreate` now also accepts, as "...", a list of such objects.
+
 # RSiena 1.3.16
 
 ## 
@@ -10,7 +24,7 @@
 
 ### Effects:
     `inPopIntnX`, `inActIntnX`, `outPopIntnX`, `outActIntnX`, `sameXInPopIntn`, 
-    `sameXOutPopIntn`, sameXInActIntn`, `sameXOutActIntn` restored
+    `sameXOutPopIntn`, `sameXInActIntn`, `sameXOutActIntn` restored
     (these had got lost in some way...).
 ### Updates:
   * All occurrences of `http` in `R` and `Rd` files changed to `https`.

R/sienaDataCreate.r

@@ -235,7 +235,22 @@ sienaDataCreate<- function(..., nodeSets=NULL, getDocumentation=FALSE)
 	dots <- as.list(substitute(list(...)))[-1] ##first entry is the word 'list'
 	if (length(dots) == 0)
 	{
-		stop('need some networks')
+		stop('need some objects')
+	}
+	if (length(dots) == 1)
+	{
+		ldots <- list(...)
+		dotsIsList <- (is.list(ldots[[1]]))
+# If dotsIsList, it needs to be a list of variables
+		if (dotsIsList) 
+		{
+			dots <- as.list(substitute(...))[-1]
+			narg <- length(ldots)
+		}
+	}
+	else
+	{
+		dotsIsList <- FALSE
 	}
 	nm <- names(dots)
 	if (is.null(nm))
@@ -255,7 +270,14 @@ sienaDataCreate<- function(..., nodeSets=NULL, getDocumentation=FALSE)
 	{
 		nm[fixup] <- dep
 	}
-	dots <- list(...)
+	if (!dotsIsList)
+	{
+		dots <- list(...)
+	}
+	else
+	{
+		dots <- (...)
+	}
 	names(dots) <- nm
 	if (any(duplicated(nm)))
 	{

R/sienaGOF.r

@@ -17,7 +17,8 @@ sienaGOF <- function(
 		sienaFitObject,	auxiliaryFunction,
 		period=NULL, verbose=FALSE, join=TRUE, twoTailed=FALSE,
 		cluster=NULL, robust=FALSE,
-		groupName="Data1", varName, tested=NULL, ...)
+		groupName="Data1", varName, tested=NULL,
+		giveNAWarning=TRUE, ...)
 	{
 	## require(MASS)
 	## require(Matrix)
@@ -88,7 +89,7 @@ sienaGOF <- function(
 		period <- 1:(attr(sienaFitObject$f[[1]]$depvars[[1]], "netdims")[3] - 1)
 	}
 
-	 obsStatsByPeriod <- lapply(period, function (j) {
+	obsStatsByPeriod <- lapply(period, function (j) {
 						matrix(
 						auxiliaryFunction(NULL,
 								sienaFitObject$f,
@@ -170,6 +171,18 @@ sienaGOF <- function(
 	  )
 	}
 
+	## Give a warning in case of missings.
+	nmissings <- vapply(simStatsByPeriod,
+		function(sp){apply(sp, 2, function(x){sum(is.na(x))})},
+					FUN.VALUE=rep(0,dim(simStatsByPeriod[[1]])[2]))
+	rownames(nmissings) <- plotKey
+	if ((sum(nmissings) > 0) & giveNAWarning)
+	{
+		cat("Number of missing values in the simulated functions:\n")
+		print(t(nmissings))
+		warning("Some simulated values are missing.")
+	}
+
 	## Aggregate by period if necessary to produce simStats
 	if (join)
 	{
@@ -213,11 +226,12 @@ sienaGOF <- function(
 		}
 		else
 		{
-			a <- cov(simulated)
+			a <- cov(simulated, use="pairwise.complete.obs")
+			a[is.na(a)] <- 0
 		}
 		ainv <- ginv(a)
 		arank <- rankMatrix(a)
-		expectation <- colMeans(simulated);
+		expectation <- colMeans(simulated)
 		centeredSimulations <- scale(simulated, scale=FALSE)
 		if (variates==1)
 		{
@@ -285,8 +299,10 @@ sienaGOF <- function(
 	}
 	else
 	{
-		covInvByPeriod <- lapply(period, function(i) ginv(
-							cov(simStatsByPeriod[[i]]) ))
+		covInvByPeriod <- lapply(period, function(i){
+				b <- cov(simStatsByPeriod[[i]], use="pairwise.complete.obs")
+				b[is.na(b)] <- 0
+				ginv(b)})
 	}
 
 	obsMhd <- sapply(period, function (i) {
@@ -327,7 +343,7 @@ sienaGOF <- function(
 					t(sienaFitObject$targets2[effectsToInclude, , drop=FALSE])
 			G <- sienaFitObject$sf2[, , effectsToInclude, drop=FALSE] -
 					rep(obsSuffStats, each=nSims)
-			sigma <- cov(apply(G, c(1, 3), sum))
+			sigma <- cov(apply(G, c(1, 3), sum), use="pairwise.complete.obs")
 			SF <- sienaFitObject$ssc[ , , effectsToInclude, drop=FALSE]
 			dimnames(SF)[[3]] <- effectsObject$effectName[effectsToInclude]
 			dimnames(G) <- dimnames(SF)
@@ -440,6 +456,7 @@ sienaGOF <- function(
 	attr(res, "simTime") <- attr(simStats,"time")
 	attr(res, "twoTailed") <- twoTailed
 	attr(res, "joined") <- join
+	attr(res, "nmissings") <- nmissings
 	res
 }
 
@@ -509,6 +526,11 @@ print.sienaGOF <- function (x, ...) {
 summary.sienaGOF <- function(object, ...) {
 	x <- object
 	print(x)
+	if (sum(attr(x, "nmissings"))> 0){
+		cat("\nThere were missing values in the simulated statistics.\n")
+		cat("Their number (by period):\n")
+		print(t(attr(x, "nmissings")))
+	}
 	if (attr(x, "scoreTest")) {
 		oneStepSpecs <- attr(x, "oneStepSpecs")
 		oneStepMhd <- attr(x, "oneStepMahalanobisDistances")

docs/manual/RSiena_Manual.tex

@@ -191,7 +191,7 @@
 by Mark Huisman, Michael Schweinberger, and Christian Steglich.
 
 This manual is frequently updated, mostly only in a minor way.
-This version was renewed for \RS version 1.3.15.
+This version was renewed for \RS version 1.3.17.
 \end{abstract}
 
 
@@ -1129,7 +1129,7 @@ \subsection{Getting help with problems}
   of \rs. Details of the latest version available can
   be found at the `Downloads' page.
   The version is
-  identified by a version number (e.g.\ 1.3.15).
+  identified by a version number (e.g.\ 1.3.17).
   You can find the number of your current installed version by
   opening \R, and typing\\
   \verb|packageVersion("RSiena")|   \\
@@ -15575,12 +15575,12 @@ \section{For programmers: Building, installing and checking the package}
 them to the repository.
 \item [Check] Checking is a process designed to ensure that packages are likely
   to work correctly when installed. Type\\
-\verb|R CMD check RSiena_1.3.15.tar.gz|\\
+\verb|R CMD check RSiena_1.3.17.tar.gz|\\
 (where the version number is adjusted to match the tar ball name.)
 \item[zip file] To make a zip file that can be used in Windows for
 `installing from a local zip file', and therefore is easy for distribution
 to others, type\\
-\verb|R CMD INSTALL --build RSiena_1.3.15.tar.gz| \\
+\verb|R CMD INSTALL --build RSiena_1.3.17.tar.gz| \\
 (where again the version number is adjusted to match the tar ball name.)
 \end{description}
 
@@ -16627,6 +16627,27 @@ \section{Changes compared to earlier versions}
 \fi
 
 
+\item 2023-01-06 GitHub, package version 1.3.17
+
+Changes in RSiena (including those for version 1.3.16):
+\begin{itemize}
+\item Effects:
+    \begin{itemize}
+    \item  \texttt{inPopIntnX},  \texttt{inActIntnX},  \texttt{outPopIntnX},  
+     \texttt{outActIntnX},  \texttt{sameXInPopIntn},\\
+     \texttt{sameXOutPopIntn}, \texttt{sameXInActIntn},  \texttt{sameXOutActIntn} restored
+    (these had got lost in some way...).
+    \end{itemize}
+\item Improvements of functionality:
+    \begin{itemize}
+    \item  \sfn{sienaGOF} now accepts simulated auxiliary statistics containing missing
+        values. If there are any, this will be reported with a warning
+        if  \texttt{giveNAWarning} is  \texttt{TRUE}.
+    \item \sfn{sienaDataCreate} now also accepts, as "...", a list of such objects.
+    \end{itemize}
+\end{itemize}
+
+
 \item 2022-11-27 GitHub, package version 1.3.15
 
 Changes in RSiena:

man/RSiena-package.Rd

@@ -44,8 +44,8 @@ see \url{https://github.com/snlab-nl/rsiena/ }.
   \tabular{ll}{
     Package: \tab RSiena\cr
     Type: \tab Package\cr
-    Version: \tab 1.3.16\cr
-    Date: \tab 2023-01-02\cr
+    Version: \tab 1.3.17\cr
+    Date: \tab 2023-01-06\cr
     Depends: \tab R (>= 3.5.0)\cr
     Imports: \tab Matrix, lattice, parallel, MASS, methods, xtable\cr
     Suggests: \tab network, tools, codetools, tcltk\cr

man/sienaDataCreate.Rd

@@ -4,15 +4,17 @@
 \title{Function to create a Siena data object}
 \description{
 Creates a Siena data object from input dependent variables
-(networks and possibly behavioural variables), covariates, and
-composition change objects.}
+    (networks and possibly behavioural variables), covariates, and
+    composition change objects.}
 \usage{
 sienaDataCreate(..., nodeSets=NULL, getDocumentation=FALSE)
 }
 \arguments{
   \item{\dots}{objects of class \code{\link{sienaDependent}},
     \code{\link{coCovar}}, \code{\link{varCovar}}, \code{\link{coDyadCovar}},
-  \code{\link{varDyadCovar}}, and/or \code{\link{sienaCompositionChange}}.
+  \code{\link{varDyadCovar}}, and/or \code{\link{sienaCompositionChange}};
+  or a list of such objects, of which the first element must not be
+  a \code{\link{sienaCompositionChange}} object.
   There should be at least one \code{\link{sienaDependent}} object.}
   \item{nodeSets}{list of Siena node sets. Default is the single node
     set named "Actors", length equal to the number of rows in the first
@@ -23,15 +25,15 @@ sienaDataCreate(..., nodeSets=NULL, getDocumentation=FALSE)
     functions, not for use by users.}
 }
 \details{
-The function checks that the objects fit, that there is at least one network,
-and adds various attributes to each dependent variable describing the
+The function checks that the objects fit, that there is at least one dependent
+variable, and adds various attributes to each dependent variable describing the
 data. If there is more than one nodeSet they must all be specified.
 Function \code{\link{print01Report}} will give a basic description of the data
 object and is a check useful, e.g., for diagnosing problems.
 }
 \value{
 An object of class "siena" which is designed to be used in a siena
-model fit. The components of the object are.
+model fit by \code{link{siena07}}. The components of the object are:
 \item{nodeSets}{List of node sets involved}
 \item{observations}{Integer indicating number of waves of data}
 \item{depvars}{List of networks and behavior variables}
@@ -55,6 +57,8 @@ model fit. The components of the object are.
 mynet <- sienaDependent(array(c(s501, s502, s503), dim=c(50, 50, 3)))
 mybeh <- sienaDependent(s50a, type="behavior")
 mydata <- sienaDataCreate(mynet, mybeh)
+# This gives the same result as
+mydata <- sienaDataCreate(list(mynet, mybeh))
 ## And for a two-mode network
 mynet1 <- sienaDependent(array(c(s501, s502), dim=c(50, 50, 2)), nodeSet="senders")
 senders <- sienaNodeSet(50, nodeSetName="senders")

man/sienaGOF.Rd

@@ -24,7 +24,7 @@
 sienaGOF(sienaFitObject, auxiliaryFunction,
         period=NULL, verbose=FALSE, join=TRUE, twoTailed=FALSE,
         cluster=NULL, robust=FALSE, groupName="Data1",
-        varName, tested=NULL, \dots)
+        varName, tested=NULL, giveNAWarning=TRUE, \dots)
 \method{plot}{sienaGOF}(x, center=FALSE, scale=FALSE, violin=TRUE, key=NULL,
         perc=.05, period=1, position=4, fontsize=12, \dots)
 descriptives.sienaGOF(x, center=FALSE, scale=FALSE, perc=.05, key=NULL,
@@ -55,29 +55,32 @@ descriptives.sienaGOF(x, center=FALSE, scale=FALSE, perc=.05, key=NULL,
        peace of mind to the user because calculations can take some time.}
   \item{join}{Boolean: should sienaGOF do tests on all of the periods
        individually (\code{FALSE}), or sum across periods (\code{TRUE})?}
-  \item{twoTailed}{Whether to use two tails for calculating \eqn{p}-values on the
-        Monte Carlo test. Recommended for advanced users only,
+  \item{twoTailed}{Whether to use two tails for calculating \eqn{p}-values on
+        the Monte Carlo test. Recommended for advanced users only,
         as it is probably only applicable in rare cases.}
   \item{cluster}{Optionally, a \code{parallel} or \code{snow} cluster to execute
         the auxiliary function calculations on.}
   \item{robust}{ Whether to use robust estimation of the covariance matrix.}
   \item{groupName}{ Name of group; relevant for multi-group data sets.}
   \item{varName}{ Name of dependent variable.}
-  \item{tested}{ A logical vector of length \code{sienaFitObject$pp} (number of parameters),
-          indicating a subset of tested parameters;
+  \item{tested}{ A logical vector of length \code{sienaFitObject$pp}
+          (number of parameters), indicating a subset of tested parameters;
           or \code{NULL}, indicating all tested parameters; or \code{FALSE},
           indicating nothing is to be tested.}
+  \item{giveNAWarning}{If \code{TRUE}, a warning is given if any simulated
+        values are missing.}
   \item{x}{ Result from a call to sienaGOF. }
   \item{center}{ Whether to center the statistics by median during plotting.}
   \item{scale}{ Whether to scale the statistics by range during plotting.
-                scale=TRUE makes little sense without also center=TRUE.}
+                scale=\code{TRUE} makes little sense without also 
+                center=\code{TRUE}.}
   \item{violin}{ Use violin plots (vs. box plots only)? }
   \item{key}{ Keys in the plot for the levels of the auxiliary statistic
         (as given by parameter \code{levls} in the examples). }
   \item{perc}{ 1 minus confidence level for the confidence bands (two sided). }
   \item{position}{ Position where the observed value is plotted: 1=under,
-        2=to the left, 3=above, 4=to the right of the red dot. Can be a single number
-        from 1 to 4, or a vector with positions for each statistic
+        2=to the left, 3=above, 4=to the right of the red dot. Can be a single
+        number from 1 to 4, or a vector with positions for each statistic
         (possibly recycled).}
   \item{fontsize}{ Font size for the observed values plotted.}
   \item{\dots}{Other arguments; for \code{sienaGOF()}, e.g., \code{levls} as a
@@ -86,7 +89,8 @@ descriptives.sienaGOF(x, center=FALSE, scale=FALSE, perc=.05, key=NULL,
        for \code{plot.sienaGOF()}, e.g., the usual plotting parameters
        \code{main}, \code{xlab}, \code{ylab}, \code{cex}, \code{cex.main},
        \code{cex.lab}, and \code{cex.axis}.}
-  \item{showAll}{If FALSE, drops statistics with variance 0, like in the plot.}
+  \item{showAll}{If \code{FALSE}, drops statistics with variance 0, 
+       like in the plot.}
 }
 \details{
  This function is used to assess the goodness of fit of an estimated stochastic
@@ -95,10 +99,10 @@ descriptives.sienaGOF(x, center=FALSE, scale=FALSE, perc=.05, key=NULL,
  of the periods, with the simulated values for the ends of the periods.
  The differences are assessed by combining the components of the auxiliary
  statistic using the Mahalanobis distance.
- 
+
  For \code{sienaFitObject}s that were made for a multi-group data set,
  if you are not sure about the \code{groupName}s to use, these can be retrieved
- by the command \code{'names(dataObject)'} (where \code{dataObject} is 
+ by the command \code{"names(dataObject)"} (where \code{dataObject} is
  the data used to produce the \code{sienaFitObject}).
  Mostly they are \code{"Data1", "Data2"}, etc.
 

man/simstats0c.Rd

@@ -6,7 +6,7 @@
 \title{Versions of FRAN}
 \description{
  The functions to be called as "FRAN" by \code{\link{siena07}}.
- They call compiled C++.
+ They call compiled C++. Not for general users' use.
 }
 \usage{
 simstats0c(z, x, data=NULL, effects=NULL, fromFiniteDiff=FALSE,
@@ -65,7 +65,7 @@ terminateFRAN(z, x)
 \details{
   Not for general users' use.\cr
   The name of \code{simstats0c} or \code{maxlikec}
-  should be used for the element FRAN of the model object, the former  when using
+  should be used for the element FRAN of the model object, the former when using
   estimation by forward simulation, the latter for maximum likelihood estimation.
   The arguments with no defaults must be passed in on the call to
   \code{\link{siena07}}.