diff --git a/R/Lrner.R b/R/Lrner.R
index 5375cf3..ddf3ffb 100644
--- a/R/Lrner.R
+++ b/R/Lrner.R
@@ -182,8 +182,8 @@ Lrner <- R6Class("Lrner",
use_var_sel = FALSE,
verbose = TRUE) {
train_data = private$train_layer$getTrainData()
- # Train only on complete data
train_data = train_data$clone(deep = FALSE)
+ # Train only on complete data
if (private$na_rm) {
all_data = train_data$getDataFrame()
complete_data = train_data$getCompleteData()
diff --git a/R/TestMetaLayer.R b/R/TestMetaLayer.R
index 1789184..ba8ea1a 100644
--- a/R/TestMetaLayer.R
+++ b/R/TestMetaLayer.R
@@ -94,15 +94,12 @@ TestMetaLayer <- R6Class("TestMetaLayer",
#' Name of individual column IDs.
#' @param data_frame `data.frame(1)` \cr
#' \code{data.frame} of layer specific predictions.
- #' @param meta_layer `TestLayer(1)` \cr
- #' Layer where to store the [TestData] object.
#'
#' @export
# TODO: Please do not export me.
setTestData = function (id,
ind_col,
- data_frame,
- meta_layer) {
+ data_frame) {
TestData$new(id = id,
ind_col = ind_col,
data_frame = data_frame,
diff --git a/R/TrainMetaLayer.R b/R/TrainMetaLayer.R
index 0b42dd6..5c6588d 100644
--- a/R/TrainMetaLayer.R
+++ b/R/TrainMetaLayer.R
@@ -257,23 +257,26 @@ TrainMetaLayer <- R6Class("TrainMetaLayer",
#' Name of individual column IDs.
#' @param data_frame `data.frame` \cr
#' \code{data.frame} of layer specific predictions.
- #' @param target `character` \cr
- #' Name of the target variable
#'
#' @export
# TODO: Please do not export me.
setTrainData = function (id,
ind_col,
- data_frame,
- target) {
+ data_frame) {
# nocov start
if (sum(!complete.cases(data_frame)) == nrow(data_frame)) {
warning("No individual fully overlaps across all layers.")
}
# nocov end
- TrainData$new(id = id,
- data_frame = data_frame,
- train_layer = self)
+ if (self$getLrner()$getNaRm()) {
+ TrainData$new(id = id,
+ data_frame = data_frame[complete.cases(data_frame), ],
+ train_layer = self)
+ } else {
+ TrainData$new(id = id,
+ data_frame = data_frame,
+ train_layer = self)
+ }
return(self)
},
#' @description
diff --git a/R/Training.R b/R/Training.R
index 549f4ac..c11c080 100644
--- a/R/Training.R
+++ b/R/Training.R
@@ -591,7 +591,6 @@ Training <- R6Class("Training",
# This code accesses each layer (except TrainMetaLayer) level
# and get the individual IDs.
layers = layers[layers$class %in% c("TrainLayer", "TrainMetaLayer"), ]
- current_model = NULL
models = list()
for (k in layers$key) {
layer = self$getFromHashTable(key = k)
@@ -609,7 +608,6 @@ Training <- R6Class("Training",
getData = function() {
layers = self$getKeyClass()
layers = layers[layers$class %in% c("TrainLayer", "TrainMetaLayer"), ]
- current_model = NULL
all_data = list()
for (k in layers$key) {
layer = self$getFromHashTable(key = k)
diff --git a/doc/fuseMLR.html b/doc/fuseMLR.html
index 8e721e8..c177a6f 100644
--- a/doc/fuseMLR.html
+++ b/doc/fuseMLR.html
@@ -12,7 +12,7 @@
-
+
How does fuseMLR work?
@@ -340,7 +340,7 @@
How does fuseMLR work?
Cesaire Fouodo
-2024-12-11
+2024-12-12
@@ -471,7 +471,7 @@ C.1 - Creating a training
param_train_list = list(probability = TRUE,
mtry = 1L),
param_pred_list = list(),
- na_action = "na.keep")
+ na_action = "na.rm")
#> Training : training
#> Problem type : classification
#> Status : Not trained
@@ -479,7 +479,7 @@ C.1 - Creating a training
#> Layers trained : 0
#> p : 131
#> n : 50
-#> na.action : na.keep
+#> na.action : na.rm
# Create gene protein abundance layer
createTrainLayer(training = training,
train_layer_id = "proteinexpr",
@@ -502,7 +502,7 @@ C.1 - Creating a training
#> Layers trained : 0
#> p : 131 | 160
#> n : 50 | 50
-#> na.action : na.keep | na.keep
# Create methylation layer
createTrainLayer(training = training,
train_layer_id = "methylation",
@@ -525,7 +525,7 @@ C.1 - Creating a training
#> Layers trained : 0
#> p : 131 | 160 | 367
#> n : 50 | 50 | 50
-#> na.action : na.keep | na.keep | na.keep
Also add a meta-layer. We use the weighted mean (internal function to
fuseMLR) as meta-learner. Similarly learners, a
meta-learner should allow at least the arguments x and
@@ -563,7 +563,7 @@
C.1 - Creating a training
#> Layers trained : 0
#> p : 131 | 160 | 367
#> n : 50 | 50 | 50
-#> na.action : na.keep | na.keep | na.keep
+#> na.action : na.rm | na.keep | na.keep
print(training)
#> Training : training
#> Problem type : classification
@@ -572,7 +572,7 @@ C.1 - Creating a training
#> Layers trained : 0
#> p : 131 | 160 | 367
#> n : 50 | 50 | 50
-#> na.action : na.keep | na.keep | na.keep
+#> na.action : na.rm | na.keep | na.keep
Function upsetplot() is available to generate an upset
of the training data, i.e. an overview how patients overlap across
layers.
@@ -660,7 +660,7 @@ C.2 - Variable selection
#> Layers trained : 0
#> p : 19 | 1 | 35
#> n : 50 | 50 | 50
-#> na.action : na.keep | na.keep | na.keep
+#> na.action : na.rm | na.keep | na.keep
For each layer, the variable selection results show the chosen
variables.
@@ -713,8 +713,8 @@ C.2 - Train
#> Layers trained : 4
#> Var. sel. used : Yes
#> p : 19 | 1 | 35 | 3
-#> n : 50 | 50 | 50 | 69
-#> na.action : na.keep | na.keep | na.keep | na.rm
+#> n : 50 | 50 | 50 | 26
+#> na.action : na.rm | na.keep | na.keep | na.rm
We can also display a summary of training to see more
details on layer levels. Information about the training data modality,
the variable selection method and the learner stored at each layer will
@@ -729,8 +729,8 @@
C.2 - Train
#> Layers trained : 4
#> Var. sel. used : Yes
#> p : 19 | 1 | 35 | 3
-#> n : 50 | 50 | 50 | 69
-#> na.action : na.keep | na.keep | na.keep | na.rm
+#> n : 50 | 50 | 50 | 26
+#> na.action : na.rm | na.keep | na.keep | na.rm
#> ----------------
#>
#> Layer geneexpr
@@ -852,8 +852,8 @@ C.2 - Train
#> Layer : meta_layer
#> Ind. id. : IDS
#> Target : disease
-#> n : 69
-#> Missing : 43
+#> n : 26
+#> Missing : 0
#> p : 3
#> ----------------
We use extractModel() to retrieve the list of stored
@@ -875,7 +875,7 @@
C.2 - Train
#> $ geneexpr :'data.frame': 50 obs. of 21 variables:
#> $ proteinexpr:'data.frame': 50 obs. of 3 variables:
#> $ methylation:'data.frame': 50 obs. of 37 variables:
-#> $ meta_layer :'data.frame': 69 obs. of 5 variables:
+#> $ meta_layer :'data.frame': 26 obs. of 5 variables:
The three simulated training modalities and the meta-data are
returned.
diff --git a/vignettes/fuseMLR.Rmd b/vignettes/fuseMLR.Rmd
index c5b2147..b0bc08e 100644
--- a/vignettes/fuseMLR.Rmd
+++ b/vignettes/fuseMLR.Rmd
@@ -86,7 +86,7 @@ createTrainLayer(training = training,
param_train_list = list(probability = TRUE,
mtry = 1L),
param_pred_list = list(),
- na_action = "na.keep")
+ na_action = "na.rm")
```
```{r proteinexpr, include=TRUE, eval=TRUE}