Finding software metrics threshold values using ROC curves

  • Authors:

  • Raed Shatnawi;Wei Li;James Swain;Tim Newman

  • Affiliations:

  • Computer Information Systems Department, Jordan University of Science and Technology, Irbid, Jordan 22110, Jordan;Computer Science Department, The University of Alabama in Huntsville, Huntsville, AL 35899, U.S.A.;Industrial Engineering Department, The University of Alabama in Huntsville, Huntsville, AL 35899, U.S.A.;Computer Science Department, The University of Alabama in Huntsville, Huntsville, AL 35899, U.S.A.

  • Venue:
  • Journal of Software Maintenance and Evolution: Research and Practice
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

An empirical study of the relationship between object-oriented (OO) metrics and error-severity categories is presented. The focus of the study is to identify threshold values of software metrics using receiver operating characteristic curves. The study used the three releases of the Eclipse project and found threshold values for some OO metrics that separated no-error classes from classes that had high-impact errors. Although these thresholds cannot predict whether a class will definitely have errors in the future, they can provide a more scientific method to assess class error proneness and can be used by engineers easily. Copyright © 2009 John Wiley & Sons, Ltd. The focus of the study is to identify threshold values of software metrics using ROC curves. The study used the three releases of the Eclipse project and found threshold values for some object-oriented metrics that separated no-error classes from classes that had high-impact errors. Although these thresholds cannot predict whether a class will definitely have errors in the future, they can provide a more scientific method to assess class error proneness and can be used by engineers easily. Copyright © 2009 John Wiley & Sons, Ltd.