Remove #250 content

R/explain.R

@@ -19,7 +19,6 @@
 #' This will happen also if \code{verbose} is TRUE. Set both \code{verbose} and \code{precalculate} to FALSE to omit calculations.
 #' @param colorize logical. If TRUE (default) then \code{WARNINGS}, \code{ERRORS} and \code{NOTES} are colorized. Will work only in the R console.
 #' @param model_info a named list (\code{package}, \code{version}, \code{type}) containg information about model. If \code{NULL}, \code{DALEX} will seek for information on it's own.
-#' @param positive_class a character indicating by name which class should be treated as positive for binary classification (ie. which class should be associated with 1 value). Keep in mind that some models assumes positive class on its own. Then parameter will not work.
 #' @param type type of a model, either \code{classification} or \code{regression}. If not specified then \code{type} will be extracted from \code{model_info}.
 #'
 #' @return An object of the class \code{explainer}.
@@ -146,10 +145,6 @@ explain.default <- function(model, data = NULL, y = NULL, predict_function = NUL
     verbose_cat("  -> model label       : ", label, "\n", verbose = verbose)
   }
 
-  if (!is.null(positive_class)) {
-    attr(model, "positive_class") <- positive_class
-  }
-
   # REPORT: checks for data
   if (is.null(data)) {
     possible_data <- try(model.frame(model), silent = TRUE)

R/misc_yhat.R

@@ -46,13 +46,8 @@ yhat.randomForest <- function(X.model, newdata, ...) {
     # if result is a vector then ncol parameter is null
     if (is.null(ncol(pred))) return(pred)
     #  binary classification
-    if (ncol(pred) == 2) {
-      if (!is.null(attr(X.model, "positive_class"))){
-        return(unlist(pred[, attr(X.model, "positive_class")], use.names = FALSE))
-      } else {
-        return(pred[,2])
-      }
-    }
+    if (ncol(pred) == 2) return(pred[,2])
+
   } else {
     pred <- predict(X.model, newdata, ...)
   }
@@ -65,19 +60,14 @@ yhat.svm <- function(X.model, newdata, ...) {
   if (X.model$type == 0) {
     pred <- attr(predict(X.model, newdata = newdata, probability = TRUE), "probabilities")
     if (ncol(pred) == 2) { # binary classification
-      if (!is.null(attr(X.model, "positive_class"))) {
-        pred <- pred[,attr(X.model, "positive_class")]
-      } else {
-        pred <- pred[,2]
-      }
+      pred <- pred[,2]
     }
   } else {
     pred <- predict(X.model, newdata, ...)
   }
   pred
 }
 
-
 #' @rdname yhat
 #' @export
 yhat.gbm <- function(X.model, newdata, ...) {
@@ -133,7 +123,7 @@ yhat.glmnet <- function(X.model, newdata, ...) {
     if(length(dim(pred)) > 2){
       return(pred[,,1])
     }
-  # For binary classifiaction matrix with one column is returned
+    # For binary classifiaction matrix with one column is returned
     if (ncol(pred) == 1) {
       return(as.numeric(pred))
     }
@@ -159,13 +149,7 @@ yhat.ranger <- function(X.model, newdata, ...) {
     # if result is a vector then ncol parameter is null
     if (is.null(ncol(pred))) return(pred)
     # binary classification
-    if (ncol(pred) == 2 & is.null(attr(X.model, "positive_class"))) {
-      return(pred[, 2])
-    } else if (ncol(pred) == 2 & !is.null(attr(X.model, "positive_class"))) {
-      # If model was fitted with numerical 0 and 1 values matrix will not have colnames. We have to add them
-      if (is.null(colnames(pred))) colnames(pred) <- 0:(ncol(pred)-1)
-      return(pred[, attr(X.model, "positive_class")])
-    }
+    if (ncol(pred) == 2) return(pred[,2])
   }
   pred
 }
@@ -176,10 +160,8 @@ yhat.model_fit <- function(X.model, newdata, ...) {
   if (X.model$spec$mode == "classification") {
     response <- as.matrix(predict(X.model, newdata, type = "prob"))
     colnames(response) <- X.model$lvl
-    if (ncol(response) == 2 & is.null(attr(X.model, "positive_class"))) {
+    if (ncol(response) == 2) {
       response <- response[,2]
-    } else if (ncol(response) == 2 & !is.null(attr(X.model, "positive_class"))) {
-      response <- response[,attr(X.model, "positive_class")]
     }
   }
   if (X.model$spec$mode == "regression") {
@@ -194,10 +176,8 @@ yhat.model_fit <- function(X.model, newdata, ...) {
 yhat.train <- function(X.model, newdata, ...) {
   if (X.model$modelType == "Classification") {
     response <- predict(X.model, newdata = newdata, type = "prob")
-    if (ncol(response) == 2 & is.null(attr(X.model, "positive_class"))) {
+    if (ncol(response) == 2) {
       response <- response[,2]
-    } else if (ncol(response) == 2 & !is.null(attr(X.model, "positive_class"))) {
-      response <- response[,attr(X.model, "positive_class")]
     }
   }
   if (X.model$modelType == "Regression") {
@@ -222,10 +202,8 @@ yhat.lrm <- function(X.model, newdata, ...) {
 yhat.rpart <- function(X.model, newdata, ...) {
   response <- predict(X.model, newdata = newdata)
   if (!is.null(dim(response))) {
-    if (ncol(response) == 2 & is.null(attr(X.model, "positive_class"))) {
+    if (ncol(response) == 2) {
       response <- response[,2]
-    } else if (ncol(response) == 2 & !is.null(attr(X.model, "positive_class"))) {
-      response <- response[,attr(X.model, "positive_class")]
     }
   }
   response
@@ -261,3 +239,4 @@ yhat.default <- function(X.model, newdata, ...) {
 #   newdata_pool <- catboost::catboost.load_pool(newdata)
 #   catboost::catboost.predict(X.model, newdata_pool)
 # }
+