Laurae (version 0.0.0.9001)

xgboard.eval.error: Xgboard Metric Evaluation Error (Binary Accuracy)

Description

This function is a custom metric for the logging of the (binary) Accuracy.

Usage

xgboard.eval.error(preds, dtrain, dump)

Arguments

preds

Type: numeric. The predictions.

dtrain

Type: xgb.DMatrix. The training data.

dump

Type: environment. An environment created by xgboard.init.

Value

The maximum accuracy for binary data.

Examples

## Not run: ------------------------------------
# # First, we must load libraries: xgboost, data.table, and R.utils
# library(xgboost)
# library(data.table)
# library(R.utils)
# 
# # Second, we load some data
# data(agaricus.train, package='xgboost')
# data(agaricus.test, package='xgboost')
# 
# # Third, we create the xgb.DMatrices and the watchlist
# dtrain <- xgb.DMatrix(agaricus.train$data, label = agaricus.train$label)
# dtest <- xgb.DMatrix(agaricus.test$data, label = agaricus.test$label)
# watchlist <- list(train = dtrain, eval = dtest)
# 
# # Fourth, we prepare environment for Accuracy/Threshold logging on Train/Test
# # Stored in D:/debug/log.txt
# my_envir <- xgboard.init(what = c("Accuracy", "Threshold"),
#                          watchnames = c("Train", "Test"),
#                          maximizer = c(TRUE, TRUE),
#                          log = "D:/debug/log.txt")
# 
# # Fifth we spawn the xgboard to open in browser
# xgboard.run(my_envir)
# 
# # Fifth, the model is set for training using these parameters
# # Take note of eval_metric needing xgboard.xgb(f = your metric, dumper = envir)
# param <- list(max_depth = 2,
#               eta = 0.05,
#               silent = 1,
#               nthread = 2, 
#               objective = "binary:logistic",
#               eval_metric = xgboard.xgb(f = xgboard.eval.error, dumper = my_envir))
# 
# # Sixth, we train a model with full logging
# # We can notice it will update in real time
# # The number of warning messages = number of file locks which xgboost waits
# # because the log file is LOCKED when read by Xgboard (to avoid crashes)
# set.seed(0)
# bst <- xgb.train(param,
#                  dtrain,
#                  nrounds = 500,
#                  watchlist)
# 
# # If you intend to run again xgboost, you have to do the following:
# # - Reset the dump environment using xgboard.init
# # - Reset the eval_metric from the parameters (because it will use the previous envir!)
# # - If you are using an interactive console, use xgboard.time before setting the seed!
## ---------------------------------------------