Merge pull request #51 from ymattu/fix/xgb

Fix/xgb

Author: ymattu

.Rbuildignore

@@ -10,3 +10,4 @@ data-raw/
 ^README-.*\.png$
 ^logo\.png$
 ^docs$
+^cran-comments\.md$

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: MlBayesOpt
 Type: Package
 Title: Hyperparameter Tuning for Machine Learning, Using Bayesian Optimization
-Version: 0.3.2
+Version: 0.3.3
 Authors@R: person("Yuya", "Matsumura", role = c("aut", "cre"), email = "mattu.yuya@gmail.com")
 Maintainer: Yuya Matsumura <mattu.yuya@gmail.com>
 Description: This package provides the hyperparameter tuning using bayesian

NEWS.md

@@ -1,3 +1,6 @@
+# MlBayesOpt 0.3.3
+- Fix Buggs for `xgb_cv_opt()`
+
 # MlBayesOpt 0.3.2
 - Add a vignette
 - Changed a option `num_tree_range` in `rf_opt()` to `num_tree`

R/xgb_cv_opt.R

@@ -13,9 +13,14 @@
 ##'     \item \code{multi:softprob} same as softmax, but prediction outputs a vector of ndata * nclass elements, which can be further reshaped to ndata, nclass matrix. The result contains predicted probabilities of each data point belonging to each class.
 ##'     \item \code{rank:pairwise} set xgboost to do ranking task by minimizing the pairwise loss.
 ##'   }
-##' @param evalmetric evaluation metrics for validation data. Users can pass a self-defined function to it.
-##' Default: metric will be assigned according to objective(rmse for regression, and error for classification, mean average precision for ranking).
-##' List is provided in detail section.
+##' @param evalmetric evaluation metrics for validation data. Users can pass a self-defined function to it. Default: metric will be assigned according to objective(rmse for regression, and error for classification, mean average precision for ranking).
+##' \itemize{
+##'   \item \code{error} binary classification error rate
+##'   \item \code{rmse} Rooted mean square error
+##'   \item \code{logloss} negative log-likelihood function
+##'   \item \code{auc} Area under curve
+##'   \item \code{merror} Exact matching error, used to evaluate multi-class classification
+##' }
 ##' @param n_folds K for cross Validation
 ##' @param eta_range The range of eta(default is c(0.1, 1L))
 ##' @param max_depth_range The range of max_depth(default is c(4L, 6L))
@@ -114,7 +119,7 @@ xgb_cv_opt <- function(data,
   }
 
   #about classes
-  if (objectfun == "binary:logistic"){
+  if (grepl("logi", objectfun) == TRUE){
     xgb_cv <- function(object_fun,
                        eval_met,
                        num_classes,
@@ -142,7 +147,7 @@ xgb_cv_opt <- function(data,
                    early_stopping_rounds = 5, maximize = TRUE, verbose = 0,
                    nrounds = nrounds_opt)
 
-      if (eval_met %in% c("auc", "logloss", "mlogloss")) {
+      if (eval_met %in% c("auc", "ndcg", "map")) {
         s <- max(cv$evaluation_log[, 4])
       } else {
         s <- max(-(cv$evaluation_log[, 4]))
@@ -181,7 +186,7 @@ xgb_cv_opt <- function(data,
                    early_stopping_rounds = 5, maximize = TRUE, verbose = 0,
                    nrounds = nrounds_opt)
 
-      if (eval_met %in% c("auc", "logloss", "mlogloss")) {
+      if (eval_met %in% c("auc", "ndcg", "map")) {
         s <- max(cv$evaluation_log[, 4])
       } else {
         s <- max(-(cv$evaluation_log[, 4]))

R/xgb_opt.R

@@ -15,7 +15,14 @@
 ##'     \item \code{multi:softprob} same as softmax, but prediction outputs a vector of ndata * nclass elements, which can be further reshaped to ndata, nclass matrix. The result contains predicted probabilities of each data point belonging to each class.
 ##'     \item \code{rank:pairwise} set xgboost to do ranking task by minimizing the pairwise loss.
 ##'   }
-##' @param evalmetric evaluation metrics for validation data. Users can pass a self-defined function to it. Default: metric will be assigned according to objective(rmse for regression, and error for classification, mean average precision for ranking). List is provided in detail section.
+##' @param evalmetric evaluation metrics for validation data. Users can pass a self-defined function to it. Default: metric will be assigned according to objective(rmse for regression, and error for classification, mean average precision for ranking).
+##' \itemize{
+##'   \item \code{error} binary classification error rate
+##'   \item \code{rmse} Rooted mean square error
+##'   \item \code{logloss} negative log-likelihood function
+##'   \item \code{auc} Area under curve
+##'   \item \code{merror} Exact matching error, used to evaluate multi-class classification
+##' }
 ##' @param eta_range The range of eta
 ##' @param max_depth_range The range of max_depth
 ##' @param nrounds_range The range of nrounds
@@ -110,7 +117,7 @@ xgb_opt <- function(train_data,
 
 
   #about classes
-  if (objectfun == "binary:logistic"){
+  if (grepl("logi", objectfun) == TRUE){
     xgb_holdout <- function(object_fun,
                             eval_met,
                             num_classes,

README.md

@@ -60,17 +60,17 @@ res0 <- svm_cv_opt(data = iris,
                    n_folds = 3,
                    init_points = 10,
                    n_iter = 1)
-#> elapsed = 0.07   Round = 1   gamma_opt = 3.3299  cost_opt = 11.7670  Value = 0.9333 
-#> elapsed = 0.34   Round = 2   gamma_opt = 5.5515  cost_opt = 76.1740  Value = 0.9067 
-#> elapsed = 0.02   Round = 3   gamma_opt = 3.2744  cost_opt = 14.1882  Value = 0.9400 
+#> elapsed = 0.04   Round = 1   gamma_opt = 3.3299  cost_opt = 11.7670  Value = 0.9333 
+#> elapsed = 0.16   Round = 2   gamma_opt = 5.5515  cost_opt = 76.1740  Value = 0.9067 
+#> elapsed = 0.01   Round = 3   gamma_opt = 3.2744  cost_opt = 14.1882  Value = 0.9400 
 #> elapsed = 0.01   Round = 4   gamma_opt = 2.1175  cost_opt = 76.6932  Value = 0.9200 
 #> elapsed = 0.01   Round = 5   gamma_opt = 3.1619  cost_opt = 84.2154  Value = 0.9600 
-#> elapsed = 0.02   Round = 6   gamma_opt = 9.4727  cost_opt = 77.6772  Value = 0.8933 
-#> elapsed = 0.05   Round = 7   gamma_opt = 6.6175  cost_opt = 13.3914  Value = 0.9267 
+#> elapsed = 0.01   Round = 6   gamma_opt = 9.4727  cost_opt = 77.6772  Value = 0.8933 
+#> elapsed = 0.01   Round = 7   gamma_opt = 6.6175  cost_opt = 13.3914  Value = 0.9267 
 #> elapsed = 0.01   Round = 8   gamma_opt = 8.8943  cost_opt = 80.5955  Value = 0.8733 
-#> elapsed = 0.02   Round = 9   gamma_opt = 3.3808  cost_opt = 89.6793  Value = 0.9333 
-#> elapsed = 0.03   Round = 10  gamma_opt = 4.3481  cost_opt = 92.6987  Value = 0.9000 
-#> elapsed = 0.02   Round = 11  gamma_opt = 2.9508  cost_opt = 84.8600  Value = 0.9467 
+#> elapsed = 0.01   Round = 9   gamma_opt = 3.3808  cost_opt = 89.6793  Value = 0.9333 
+#> elapsed = 0.01   Round = 10  gamma_opt = 4.3481  cost_opt = 92.6987  Value = 0.9000 
+#> elapsed = 0.01   Round = 11  gamma_opt = 2.9508  cost_opt = 84.8600  Value = 0.9467 
 #> 
 #>  Best Parameters Found: 
 #> Round = 5    gamma_opt = 3.1619  cost_opt = 84.2154  Value = 0.9600