classifier

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: ngboostR
 Title: R Wrapper for NGBoost python module
-Version: 0.1.5
+Version: 0.1.6
 Authors@R: 
     person(given = "Alfonso",
            family = "Carabantes Alamo",

R/distributions.R

@@ -79,7 +79,7 @@ k_categorical <- function( K ){
 #' @return Bernoulli Distribution Object
 #' @export
 Bernoulli <- function(){
-  ngboost$distns$k_categorical(as.integer(1))
+  ngboost$distns$k_categorical(as.integer(2))
 }
 
 

R/ngboostRClassifier.R

@@ -11,24 +11,29 @@
 #' @param tol numerical tolerance to be used in optimization
 #' @return A NGBoostRegressor object
 #' @export
-create_classifier <- function(Dist=NULL,
-                              natural_gradient=TRUE,
-                              n_estimators=as.integer(500),
-                              learning_rate=0.01,
-                              minibatch_frac=1.0,
+create_classifier <- function(Base=DecisionTreeRegressor(),
+                              Dist=Bernoulli(),
                               col_sample=1.0,
-                              verbose=TRUE,
-                              verbose_eval=as.integer(100),
-                              tol=1e-4) {
+                              learning_rate=0.01, 
+                              minibatch_frac=1.0, 
+                              n_estimators=as.integer(500),
+                              natural_gradient=TRUE,
+                              random_state=NULL, 
+                              tol=0.0001,
+                              verbose=TRUE, 
+                              verbose_eval=as.integer(100)) {
 
-  classifier <- ngboost$NGBClassifier(Dist=Dist,
-                                      natural_gradient=natural_gradient,
-                                      n_estimators=as.integer(n_estimators),
-                                      minibatch_frac=minibatch_frac,
+  classifier <- ngboost$NGBClassifier(Base=Base,
+                                      Dist=Dist,
                                       col_sample=col_sample,
-                                      verbose=verbose,
-                                      verbose_eval=as.integer(verbose_eval),
-                                      tol=tol
+                                      learning_rate=learning_rate, 
+                                      minibatch_frac=minibatch_frac, 
+                                      n_estimators=n_estimators,
+                                      natural_gradient=natural_gradient,
+                                      random_state=random_state, 
+                                      tol=tol,
+                                      verbose=verbose, 
+                                      verbose_eval=verbose_eval
                                       )
   classifier
 }
@@ -77,6 +82,7 @@ predict_classifier_prob <- function( ngbr_cla, new_data) {
 #' @export
 predict_classifier_dist <- function( ngbr_cla, new_data) {
   pred_temp <- ngbr_cla$pred_dist(new_data)
-  pred <- list( "p0"=pred_temp$loc, "p1"=pred_temp$scale)
+  number_classes <- ngbr_cla$Dist$K_
+  pred <- pred_temp$params
   pred
 }

vignettes/classification.Rmd

@@ -0,0 +1,55 @@
+---
+title: "ngboostR Classifier"
+author: "Alfonso Carabantes"
+date: "22/8/2021"
+output: html_document
+---
+
+```{r}
+#Install last ve
+#library(devtools)
+#install_github("acca3003/ngboostR")
+
+library(ngboostR) # R implementation for NGBoost
+library(Metrics) # Métrics
+library(mlbench) # Breast Cancer dataset
+library(caret)
+
+
+data("BreastCancer")
+set.seed(999)
+BreastCancer <- na.omit(BreastCancer)
+trainIndex <- createDataPartition(BreastCancer$Class, p = .8, 
+                                  list = FALSE, 
+                                  times = 1)
+
+X_train <- BreastCancer[trainIndex,2:10]
+Y_train <- BreastCancer[trainIndex,11]
+Y_train <- as.integer(as.integer(Y_train)-1)
+
+X_val <- BreastCancer[-trainIndex,2:10]
+Y_val <- BreastCancer[-trainIndex,11]
+Y_val <- as.integer(as.integer(Y_val)-1)
+
+
+# Create the regressor object
+# reg_ngboost <- create_regressor() # Default parameters
+class_ngboost <- create_classifier()
+# Train with the boston data
+fit_classifier(class_ngboost, X_train, Y_train, X_val, Y_val)
+
+
+# Predict the price
+predictions <- predict_classifier(class_ngboost, X_val)
+Metrics::accuracy(Y_val,predictions)
+
+# Predict the price
+predictions_prob <- predict_classifier_prob(class_ngboost, X_val)
+predictions_prob
+
+# Predict the price as a distribution
+predictions_dist <- predict_classifier_dist(class_ngboost, X_val)
+predictions_dist
+
+```
+