evaluation-measures: Performance measures used in classifier evaluation

In RSSL: Implementations of Semi-Supervised Learning Approaches for Classification

Performance measures used in classifier evaluation

Description

Classification accuracy on test set and other performance measure that can be used in CrossValidationSSL and LearningCurveSSL

Usage

measure_accuracy(trained_classifier, X_l = NULL, y_l = NULL, X_u = NULL,
  y_u = NULL, X_test = NULL, y_test = NULL)

measure_error(trained_classifier, X_l = NULL, y_l = NULL, X_u = NULL,
  y_u = NULL, X_test = NULL, y_test = NULL)

measure_losstest(trained_classifier, X_l = NULL, y_l = NULL, X_u = NULL,
  y_u = NULL, X_test = NULL, y_test = NULL)

measure_losslab(trained_classifier, X_l = NULL, y_l = NULL, X_u = NULL,
  y_u = NULL, X_test = NULL, y_test = NULL)

measure_losstrain(trained_classifier, X_l = NULL, y_l = NULL, X_u = NULL,
  y_u = NULL, X_test = NULL, y_test = NULL)

Arguments

trained_classifier	the trained classifier object
X_l	design matrix with labeled object
y_l	labels of labeled objects
X_u	design matrix with unlabeled object
y_u	labels of unlabeled objects
X_test	design matrix with test object
y_test	labels of test objects

Functions

• measure_error(): Classification error on test set

• measure_losstest(): Average Loss on test objects

• measure_losslab(): Average loss on labeled objects

• measure_losstrain(): Average loss on labeled and unlabeled objects

See Also

Other RSSL utilities: LearningCurveSSL(), SSLDataFrameToMatrices(), add_missinglabels_mar(), df_to_matrices(), missing_labels(), split_dataset_ssl(), split_random(), true_labels()

RSSL documentation built on May 29, 2024, 2:38 a.m.