search for


Amended agreement chart for the confusion matrix
Journal of the Korean Data & Information Science Society 2022;33:551-65
Published online July 31, 2022;
© 2022 Korean Data and Information Science Society.

Chong Sun Hong1 · Ye Won Choi2

12Department of Statistics, Sungkyunkwan University
Correspondence to: 1 Professor, Department of Statistics, Sungkyunkwan University, 25-2,
Sungkyunkwan-Ro, Jongno-Gu, Seoul, 03063, Korea. E-mail:
2 Master course student, Department of Statistics, Sungkyunkwan University, Seoul, 03063, Korea.
Received March 2, 2022; Revised March 15, 2022; Accepted March 21, 2022.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
The ROC curve, which expresses the relationship between TPR (true positive rate) and FPR (false positive rate), is a well-known and useful method of exploring the performance of a binary classification model. Among many methods to compensate for the disadvantages of the ROC curve, Bangdiwala et al. (2008) proposed the agreement chart that expresses the degree of agreement corresponding to TP (true positive) and TN (true negative) as the squared frequencies. The amended agreement chart whose horizontal and vertical axes consist with the square roots of four cells of a confusion matrix is proposed in this paper. From the amended agreement chart, it is possible to explore the frequencies and sample sizes that make up the confusion matrix whereas these cannot be expressed in the ROC curve and agreement chart. It was also found that additional information about TPR, FPR, the positive and negative predictive values, and the positive and negative likelihood ratios could be visually implemented. Characteristics and properties of the amended agreement chart are discussed in many cases classified by different sample sizes and various accuracy measures.
Keywords : Accuracy, misclassify, sensitivity, specificity, threshold.