An Analogy Based Technique for Predicting the Vector Impact of Software Change Requests
Date
2018-01-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Managing the impact of software change request (CR) requires proper prediction to minimize cost and enhance quality. Traditional techniques only identify source code entities likely to be impacted by the implementation of CR. However, looking at code entity is not enough. To ensure maximum utility and return on investment, decision-makers also need to consider effort and duration requirements. It is more challenging for Minimum Viable Products which need to implement essential CR promptly. This thesis proposes an analogy based reasoning approach called Vector Change Impact Analysis which can generate file, effort and duration predictions simultaneously. Case-studies on industrial and open source projects highlight effectiveness of VCIA. The proposed approach is integrated into an analytics dashboard, for software company Brightsquid Inc. Analyzing performance results under different settings, VCIA achieves 67% file impact prediction recall. For effort and duration prediction, achieved accuracy is 44% and 30% respectively. Applying VCIA to complement expert knowledge has resulted in improved planning and organization, higher customer satisfaction and better forecasting of future returns.
Description
Keywords
Citation
Kabeer, S. J. (2018). An Analogy Based Technique for Predicting the Vector Impact of Software Change Requests (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.