Sökning: id:"swepub:oai:gup.ub.gu.se/47929" >
Predicting Monthly ...
Predicting Monthly Defect Inflow in Large Software Projects – An Industrial Case Study
-
- Staron, Miroslaw, 1977 (författare)
- Gothenburg University,Göteborgs universitet,Institutionen för tillämpad informationsteknologi (GU),Department of Applied Information Technology (GU)
-
Meding, Wilhelm (författare)
-
(creator_code:org_t)
- 2007
- 2007
- Engelska.
-
Ingår i: Industry Track Proceedings of the 27 International Symposium on Software Reliability Engineering. ; , s. 23-30
- Relaterad länk:
-
https://gup.ub.gu.se...
Abstract
Ämnesord
Stäng
- One of the main aspects considered during planning of large software projects is the defect inflow. The curve of the defect inflow shows what potentially can happen if no counter measures are taken in the project. The curve can support project and quality managers in taking decisions, in order to meet some of the required quality objectives. This paper presents a method for constructing prediction models of a monthly defect inflow for the duration of an entire project. The projects, which we construct the models for, are large software projects at Ericsson, which are structured around a significant number of work packages, which incrementally deliver new functionality for embedded software in a network node. The presented method is based on using similarity of projects for estimations and regression techniques, which result in equations describing defect inflow in a project. The independent variable in the equation is the month of the project. The method results in creating prediction models which we evaluate in two on-going projects by comparing them to existing practices at Ericsson and alternative ways of constructing the models. The method is intended to be simple and allows adjusting the models as the project progresses. The results from these two projects show that the accuracy of predictions is higher for the models developed using our method than for such models as the Rayleigh model.
Ämnesord
- NATURVETENSKAP -- Data- och informationsvetenskap (hsv//swe)
- NATURAL SCIENCES -- Computer and Information Sciences (hsv//eng)
Nyckelord
- Software Engineering
Publikations- och innehållstyp
- ref (ämneskategori)
- kon (ämneskategori)