A Method for Predicting Defects in System Testing for V-Model

A Method for Predicting Defects in System Testing for V-Model

(Paper ID: 37)

Muhammad Dhiauddin bin Mohamed SuffianFaculty of Computer Science & Information System

mdhiauddin2@live.utm.my

AP Dr. Suhaimi bin IbrahimAdvanced Informatics School

suhaimiibrahim@utm.my

PARIS 2012

Presentation Outline

• Introduction• Related Work• Findings and Discussion• Case Study• Conclusion and Recommendation

Introduction• V-model emphasizes on early testing activities:

– rigor verification and validation activities throughout the phases: reviews, inspection, unit testing, integration testing and system testing

• System testing is important in V-model:– Carried out by independent team– Ensure that all defects are discovered within the phase – Validate that the software or system under test meets the specification

• Independent testing team faces challenges in completing test– Meeting the deadline to ensure on-time release– Finding as many defects as possible

ProblemProblem

• There is a need to have early indicator of total defects to be found in system testing There is a need to have early indicator of total defects to be found in system testing before it startsbefore it starts• Systematic method needs to be developed for predicting defects in system testing Systematic method needs to be developed for predicting defects in system testing using metrics from prior phases (development-related + testing-related activities) using metrics from prior phases (development-related + testing-related activities)

Introduction (cont.)• How does prediction help independent testing team?

Related Work

Findings and Discussion

Findings and Discussion(cont.)

Findings and Discussion(cont.)

Findings and Discussion(cont.)

Case StudyData set for regression analysis

Metrics

• Number of requirement pages• Number of design pages• Code size in a form of lines of code• Total test cases• Total effort in test case design• Total effort in phases prior to system testing• Requirement error• Design error• Code error• Test cases error• Total defects logged in a form of all defects and functional defects

Case Study (cont.)

Case Study (cont.)Verification result

Functional Defects Functional Defects = 4.00 - 0.204 Requirement Error - 0.631 Coding = 4.00 - 0.204 Requirement Error - 0.631 Coding Error + 1.90 KLOC – 0.140 Requirement Page + Error + 1.90 KLOC – 0.140 Requirement Page +

0.125 Design Page – 0.169 Total Test Cases + 0.125 Design Page – 0.169 Total Test Cases + 0.221Total Effort Days 0.221Total Effort Days

Selected equation

Conclusion and Recommendation• Achievement:

– The proposed method provides systematic way towards predicting defects for system testing in V-model by using prior phases’ metrics.

– Statistical analysis used serves as the powerful tool to measure how good the method is in determining the accuracy of the prediction

– Having maximum and minimum range for predicting defects allows independent testing team to have a control plan on what to do should the prediction does not fall within the specified range

• Future works:– More metrics are taken into consideration to construct the prediction,

particularly product-related metrics.– Future prediction could also forecast non-functional defects as well as defects

based on severity. – Having specific prediction for different types of software which makes it more

practical and useful.

THANK YOU