moving ctsem dependency to suggests

adding heuristic to choose mtry in forests
added match.arg check in control
added semforest_control() alias

DESCRIPTION

@@ -13,8 +13,8 @@ Depends:
 Imports:
     rpart,
     rpart.plot (>= 3.0.6),
-    cluster,
     lavaan,
+    cluster,
     ggplot2,
     tidyr,
     methods,
@@ -25,7 +25,6 @@ Imports:
     clisymbols,
     future.apply,
     data.table,
-    ctsemOMX,
     expm,
     gridBase
 Suggests:
@@ -35,7 +34,8 @@ Suggests:
     MASS,
     psychTools,
     testthat,
-    future
+    future,
+    ctsemOMX
 Description: SEM Trees and SEM Forests -- an extension of model-based decision
     trees and forests to Structural Equation Models (SEM). SEM trees hierarchically
     split empirical data into homogeneous groups each sharing similar data patterns

NAMESPACE

@@ -59,6 +59,7 @@ export(prune)
 export(se)
 export(semforest)
 export(semforest.control)
+export(semforest_control)
 export(semforest_score_control)
 export(semtree)
 export(semtree.constraints)

R/growTree.R

@@ -315,10 +315,6 @@ growTree <- function(model = NULL, mydata = NULL,
         return(NULL)
       }
     )
-  }
-  # 3. Traditional cross validation for covariate split selection
-  else if (control$method == "cv") {
-    stop("This split selection procedure is not supported anymore. Please see the new score-based tests for split selection.")
   } else {
     ui_fail("Error. Unknown split method selected")
     stop()

R/semforest.R

@@ -114,6 +114,13 @@ semforest <- function(model,
 
   }
 
+  # set mtry heuristically if not set manually
+  if (is.null(semforest.control$mtry)) {
+    num_covariates <- length(covariate.ids)
+    mtry <- ceil(sqrt(num_covariates))
+    ui_message("Setting mtry = ",mtry," based on ",num_covariates," predictors.\n")
+  }
+
   # pass mtry from forest to tree control
   if (!is.na(semforest.control$semtree.control$mtry)) {
     ui_stop(

R/semforest.control.R

@@ -4,7 +4,7 @@
 #' algorithm.
 #'
 #'
-#' @aliases semforest.control print.semforest.control semforest_score_control
+#' @aliases semforest.control semforest_control print.semforest.control semforest_score_control
 #' @param num.trees Number of trees.
 #' @param sampling Sampling procedure. Can be subsample or bootstrap.
 #' @param control A SEM Tree control object. Will be generated by default.
@@ -55,3 +55,9 @@ semforest_score_control <- function(...)
 
   return(ctrl)
 }
+
+#' @export
+semforest_control <- function(...)
+{
+  semforest.control(...)
+}

R/semtree.control.R

@@ -99,7 +99,7 @@
 #' 
 #' @export
 semtree.control <-
-  function(method = "naive",
+  function(method = c("naive","score","fair","fair3"),
            min.N = 20,
            max.depth = NA,
            alpha = .05,
@@ -146,7 +146,7 @@ semtree.control <-
     # minimum number of cases in leaf
     options$min.bucket <- min.bucket
     # method
-    options$method <- method
+    options$method <- match.arg(method)
     # maximal depth of the tree , set to NA for unrestricted trees
     options$max.depth <- max.depth
     # test invariance of strong restrictions

vignettes/forests.Rmd

@@ -69,7 +69,7 @@ summary(result)
 Create a forest control object that stores all tuning parameters of the forest. Note that we use only 5 trees for illustration. Please increase the number in real applications to several hundreds. To speed up computation time, consider score-based test for variable selection in the trees.
 
 ```{r}
-control <- semforest.control(num.trees = 5)
+control <- semforest_control(num.trees = 5)
 print(control)
 ```