Merge pull request #35 from mi-erasmusmc/develop

Develop 0.4.0

.Rbuildignore

@@ -1,3 +1,5 @@
+^renv$
+^renv\.lock$
 ^.*\.Rproj$
 ^\.Rproj\.user$
 ^vignettes$

.gitignore

@@ -1,55 +1,46 @@
 # History files
 .Rhistory
-
 # Session Data files
 .RData
 .RDataTmp
-
 # User-specific files
 .Ruserdata
-
 # Output files from R CMD check
 /*.Rcheck/
-
 # RStudio files
 .Rproj.user/
-
 # Produced vignettes
 vignettes/*.html
 vignettes/*.pdf
-
 # R Environment Variables
 .Renviron
-
 # Documents and results
 docs/
 output/
-
 # Mac files
 .DS_STORE
 .idea/
-
 # Compiled Object files
 *.o
 *.obj
-
 # Compiled Dynamic libraries
 *.so
 *.dylib
 *.dll
-
 # Compiled Static libraries
 *.a
 *.lib
-
 # Other C++ files
 **/__history/
-
 # EXPLORE result files
 *.result
-inst/doc
+inst/cmake-build-debug
+inst/cmake-build-debug-coverage
 docs
-
 # PaRe Report
 Report.html
 Report_files/
+renv.lock
+renv
+.Rprofile
+@OutputFile

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: Explore
 Type: Package
 Title: Exhaustive Procedure for LOgic-Rule Extraction (EXPLORE)
-Version: 0.3.0
+Version: 0.4.0
 Date: 2020-12-14
 Authors@R: c(
     person("Aniek", "Markus", , "[email protected]", role = c("aut"), comment = c(ORCID = "0000-0001-5779-4794")),
@@ -18,10 +18,12 @@ Imports:
     RcppParallel,
     stringr,
     caret,
-    pracma
+    pracma,
+    glue,
+    magrittr
 Encoding: UTF-8
 LinkingTo: Rcpp, BH (>= 1.51.0), RcppParallel
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.2
 Suggests: 
     testthat (>= 3.0.0),
     knitr,

NAMESPACE

@@ -1,14 +1,17 @@
 # Generated by roxygen2: do not edit by hand
 
+export(candidateNumberExplore)
 export(modelsCurveExplore)
 export(predictExplore)
+export(resultsExplore)
 export(rocCurveExplore)
 export(trainExplore)
 import(Rcpp)
 import(checkmate)
 importFrom(RcppParallel,RcppParallelLibs)
 importFrom(caret,confusionMatrix)
 importFrom(farff,writeARFF)
+importFrom(magrittr,"%>%")
 importFrom(pracma,trapz)
 importFrom(stringr,str_extract)
 importFrom(stringr,str_replace_all)

NEWS.md

@@ -1,3 +1,13 @@
+# Explore 0.4.0
+
+* Improved testing for different datasets. 
+
+* trainExplore has a parameter to select the result type.  
+
+* Added testExplore for test configuration. 
+
+* Added resultExplore to access results. 
+
 # Explore 0.3.0
 
 * Package tested on Windows, MacOs and Linux. R-CMD-Check without errors.

R/HelperFunctions.R

@@ -78,8 +78,13 @@ saveData <- function(output_path, train_data, file_name) {
 
   # Fix col type for binary data
   binary_cols <- sapply(1:ncol(train_data), function(c) all(train_data[[c]] %in% 0:1))
-  train_data[binary_cols] <- lapply(colnames(train_data[binary_cols]), function(c) factor(train_data[[c]], labels=c(0,1)))
+
+  # Convert TRUE/FALSE to 1/0
+  train_data <- convert_logical(train_data)
 
+  # Order data (first binary then continuous features)
+  train_data <- cbind(train_data[binary_cols],train_data[!binary_cols]) # Order needed for correct functioning of main algorithm in C++ 
+
   # Save data as arff file
   if (file.exists(paste0(output_path, file_name, ".arff"))) {file.remove(paste0(output_path, file_name, ".arff"))}
   farff::writeARFF(train_data, paste0(output_path, file_name, ".arff"))
@@ -97,9 +102,43 @@ saveData <- function(output_path, train_data, file_name) {
   # TODO: Support other file formats?
 }
 
+convert_logical <- function(train_data) {
+
+  binary_cols <- sapply(train_data, function(col) all(col %in% c(0, 1, TRUE, FALSE)))
+
+  # Convert TRUE/FALSE to 1/0 and create factors
+  train_data[binary_cols] <- lapply(train_data[binary_cols], function(col) {
+    col <- as.numeric(as.logical(col))  # Convert TRUE/FALSE to 1/0
+    factor(col, levels = c(0, 1), labels = c(0, 1))  # Convert to factors
+  })
+
+  return(train_data)
+
+}
+
 # Correlation metric for binary data.
 jaccard <- function(a, b) {
   intersection = length(intersect(a, b))
   union = length(a) + length(b) - intersection
   return (intersection/union)
-}
+}
+
+phi <- function(a, b) {
+  contingency_tb <- table(a, b)
+
+  r.sum <- rowSums(contingency_tb)
+  c.sum <- colSums(contingency_tb)
+
+  total <- sum(r.sum)
+  r.sum <- r.sum/total
+  c.sum <- c.sum/total
+
+  v <- prod(r.sum, c.sum)
+  phi <- (contingency_tb[1,1] / total - c.sum[1] * r.sum[1] / sqrt(v))
+  names(phi) <- NULL
+
+  return(phi)
+}
+
+
+