Comparative study of quality estimation of binary classification

The paper describes results of analytical and experimental analysis of seventeen functions used for evaluation of binary classification results of arbitrary data. The results are presented by 2×2 error matrices. The behavior and properties of the main functions calculated by the elements of such mat...

Full description

Saved in:
Bibliographic Details
Main Authors: V. V. Starovoitov, Yu. I. Golub
Format: Article
Language:Russian
Published: National Academy of Sciences of Belarus, the United Institute of Informatics Problems 2020-03-01
Series:Informatika
Subjects:
Online Access:https://inf.grid.by/jour/article/view/1044
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849240220504948736
author V. V. Starovoitov
Yu. I. Golub
author_facet V. V. Starovoitov
Yu. I. Golub
author_sort V. V. Starovoitov
collection DOAJ
description The paper describes results of analytical and experimental analysis of seventeen functions used for evaluation of binary classification results of arbitrary data. The results are presented by 2×2 error matrices. The behavior and properties of the main functions calculated by the elements of such matrices are studied.  Classification options with balanced and imbalanced datasets are analyzed. It is shown that there are linear dependencies between some functions, many functions are invariant to the transposition of the error matrix, which allows us to calculate the estimation without specifying the order in which their elements were written to the matrices.It has been proven that all classical measures such as Sensitivity, Specificity, Precision, Accuracy, F1, F2, GM, the Jacquard index are sensitive to the imbalance of classified data and distort estimation of smaller class objects classification errors. Sensitivity to imbalance is found in the Matthews correlation coefficient and Kohen’s kappa. It has been experimentally shown that functions such as the confusion entropy, the discriminatory power, and the diagnostic odds ratio should not be used for analysis of binary classification of imbalanced datasets. The last two functions are invariant to the imbalance of classified data, but poorly evaluate results with approximately equal common percentage of classification errors in two classes.We proved that the area under the ROC curve (AUC) and the Yuden index calculated from the binary classification confusion matrix are linearly dependent and are the best estimation functions of both balanced and imbalanced datasets.
format Article
id doaj-art-72524e5cf24f46d1a618663b8f56f96f
institution Kabale University
issn 1816-0301
language Russian
publishDate 2020-03-01
publisher National Academy of Sciences of Belarus, the United Institute of Informatics Problems
record_format Article
series Informatika
spelling doaj-art-72524e5cf24f46d1a618663b8f56f96f2025-08-20T04:00:40ZrusNational Academy of Sciences of Belarus, the United Institute of Informatics ProblemsInformatika1816-03012020-03-011718710110.37661/1816-0301-2020-17-1-87-101914Comparative study of quality estimation of binary classificationV. V. Starovoitov0Yu. I. Golub1The United Institute of Informatics Problems of the National Academy of Sciences of BelarusThe United Institute of Informatics Problems of the National Academy of Sciences of BelarusThe paper describes results of analytical and experimental analysis of seventeen functions used for evaluation of binary classification results of arbitrary data. The results are presented by 2×2 error matrices. The behavior and properties of the main functions calculated by the elements of such matrices are studied.  Classification options with balanced and imbalanced datasets are analyzed. It is shown that there are linear dependencies between some functions, many functions are invariant to the transposition of the error matrix, which allows us to calculate the estimation without specifying the order in which their elements were written to the matrices.It has been proven that all classical measures such as Sensitivity, Specificity, Precision, Accuracy, F1, F2, GM, the Jacquard index are sensitive to the imbalance of classified data and distort estimation of smaller class objects classification errors. Sensitivity to imbalance is found in the Matthews correlation coefficient and Kohen’s kappa. It has been experimentally shown that functions such as the confusion entropy, the discriminatory power, and the diagnostic odds ratio should not be used for analysis of binary classification of imbalanced datasets. The last two functions are invariant to the imbalance of classified data, but poorly evaluate results with approximately equal common percentage of classification errors in two classes.We proved that the area under the ROC curve (AUC) and the Yuden index calculated from the binary classification confusion matrix are linearly dependent and are the best estimation functions of both balanced and imbalanced datasets.https://inf.grid.by/jour/article/view/1044binary classificationconfusion matrixfunctions of accuracy classificationarea under roc curveyouden’s index
spellingShingle V. V. Starovoitov
Yu. I. Golub
Comparative study of quality estimation of binary classification
Informatika
binary classification
confusion matrix
functions of accuracy classification
area under roc curve
youden’s index
title Comparative study of quality estimation of binary classification
title_full Comparative study of quality estimation of binary classification
title_fullStr Comparative study of quality estimation of binary classification
title_full_unstemmed Comparative study of quality estimation of binary classification
title_short Comparative study of quality estimation of binary classification
title_sort comparative study of quality estimation of binary classification
topic binary classification
confusion matrix
functions of accuracy classification
area under roc curve
youden’s index
url https://inf.grid.by/jour/article/view/1044
work_keys_str_mv AT vvstarovoitov comparativestudyofqualityestimationofbinaryclassification
AT yuigolub comparativestudyofqualityestimationofbinaryclassification