stricter NAMESPACE, ggplot2 now imported too

rvw-org · Aug 22, 2016 · 30270a9 · 30270a9
1 parent 1696bec
commit 30270a9
Show file tree

Hide file tree

Showing 6 changed files with 36 additions and 15 deletions.
diff --git a/ChangeLog b/ChangeLog
@@ -1,3 +1,15 @@
+2016-08-22  Dirk Eddelbuettel  <edd@debian.org>
+
+	* DESCRIPTION (License): Use dual license with GPL (>= 2) for my code,
+	and BSD 3-clause for existing code
+
+	* NAMESPACE: All importing now via importFrom()
+
+	* demo/vw_example_4.R: Add library(pROC), add a legend to ROC plot
+	* demo/00Index: Add vw_example4
+
+	* R/plot.R (plotDensity): add utils::globalVariables() for R CMD check
+
 2016-08-21  Dirk Eddelbuettel  <edd@debian.org>
 
 	* DESCRIPTION (Suggests): Added caret (confusion matrix), ggplot2

diff --git a/DESCRIPTION b/DESCRIPTION
@@ -10,6 +10,6 @@ Description: A basic R interface to the Vowpal Wabbit fast on-line learner
  is provided.
 SystemRequirements: The Vowpal Wabbit binary 'vw' has to be in the path.
 License: GPL (>= 2) | BSD_3_clause + file LICENSE
-Imports: pROC, data.table
-Suggests: yaml, caret, ggplot2, earth
+Imports: pROC, data.table, ggplot2
+Suggests: yaml, caret, earth
 RoxygenNote: 5.0.1
diff --git a/NAMESPACE b/NAMESPACE
@@ -1,7 +1,8 @@
 
-export(dt2vw)
-export(vw)
+export("vw",
+       "dt2vw")
 
-import("data.table")
-import("pROC")
 importFrom("utils", "write.table")
+importFrom("data.table", "data.table", "setDT", "setnames", "fread", "set")
+importFrom("pROC", "roc", "plot.roc")
+importFrom("ggplot2", "ggplot", "aes", "geom_density")
diff --git a/R/dt2vw.R b/R/dt2vw.R
@@ -120,8 +120,10 @@ dt2vw <- function(data, fileName, namespaces = NULL, target, weight = NULL, tag
     }
 
     ##appending the name of the variable to its value for each categorical variable
-    sapply(Index, FUN = function(x){sapply(names(x), FUN = function(y){if(x[[y]] == F){
-                                                                           set(data, i=NULL, y, paste0(y,"_",data[[y]]))
+    sapply(Index,
+           FUN = function(x){sapply(names(x),
+                                    FUN = function(y){if(x[[y]] == F){
+                                                          set(data, i=NULL, y, paste0(y,"_",data[[y]]))
                                                     }})})
 
     ##   FIRST PART OF THE VW DATA FORMAT: target, weight, tag

diff --git a/R/plot.R b/R/plot.R
@@ -14,3 +14,6 @@ plotDensity <- function(x) {
     p <- ggplot(data=x) + geom_density(aes(x=predicted,linetype=actual,color=actual))
     p
 }
+
+## to silence R CMD check --as-cran
+utils::globalVariables(c("predicted", "actual"))
diff --git a/R/vw.R b/R/vw.R
@@ -120,7 +120,7 @@ vw <- function(training_data, validation_data,  model='mdl.vw',
     cmd <- sprintf('%s -d %s --loss_function %s -f %s', getVW(), training_data, loss, model)
     cmd <- sprintf('%s --learning_rate=%s --passes %s', cmd, learning_rate, passes)
     if (use_cache) cmd <- sprintf("%s -c", cmd)
-    
+
     if (!is.null(l1)) cmd <- sprintf('%s --l1 %s', cmd, l1)
     if (!is.null(l2)) cmd <- sprintf('%s --l2 %s', cmd, l2)
 
@@ -141,7 +141,9 @@ vw <- function(training_data, validation_data,  model='mdl.vw',
         del_prob <- FALSE
 
     validation_data <- path_data_args[[2]]
-    predict <- sprintf('%s -t -i %s -p %s %s -d %s', getVW(), model, out_probs, link_function, validation_data)
+    predict <- sprintf('%s -t -i %s -p %s %s -d %s',
+                       getVW(), model, out_probs,
+                       link_function, validation_data)
     system(predict)
 
     if (do_evaluation) {
@@ -159,13 +161,14 @@ vw <- function(training_data, validation_data,  model='mdl.vw',
 
         if (use_perf) {
             ## compute auc using perf
-            eval_model <- sprintf("%s -ROC -files %s %s | cut -c8-14", getPerf(), validation_labels, out_probs)
+            eval_model <- sprintf("%s -ROC -files %s %s | cut -c8-14",
+                                  getPerf(), validation_labels, out_probs)
             auc <- system(eval_model, intern = TRUE)
         } else {
             auc <- roc_auc(out_probs, validation_labels, plot_roc, cmd)
         }
 
-        if (verbose) 
+        if (verbose)
             cat('\nModel Parameters\n', cmd, "\n\nAUC: ", auc, "\n")
     }
 
@@ -176,21 +179,21 @@ vw <- function(training_data, validation_data,  model='mdl.vw',
         for (i in 1:2) {
             if ("data.frame" %in% class(data_args[[i]])) {
                 file.remove(path_data_args[[i]])
-                cachefile <- paste0(path_data_args[[i]], ".cache") 
+                cachefile <- paste0(path_data_args[[i]], ".cache")
                 if (file.exists(cachefile)) unlink(cachefile)
             }
         }
         if (del_prob) file.remove(out_probs)
         if (exists("del_val") && del_val) file.remove(validation_labels)
         if (file.exists("mdl.vw")) file.remove("mdl.vw")
     }
-    
+
     ##return(list(auc=auc, preds=probs))
     return(list(auc=auc,
                 data=setDT(data.frame(predicted=probs)), #, actual=as.factor(validation_data[, factor]))),
                 workingdir=getwd(),
                 cmd=cmd))
-           
+
 }
 
 # Reads labels file (from the validation dataset)