A hybrid effort estimation model for change request in ongoing software project
Loading...
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Teknologi Malaysia
Abstract
Frequent requirement changes are challenging in an ongoing software project and have been recognised as one of the main causes of project failures. Many change requests are expected as new requirements would eventually accumulate to meet the stakeholders’ expectations or adapt to new technologies and current circumstances. This circumstance requires the Software Project Manager (SPM) to make good and fast effort estimations. Inaccurate estimations could lead to over-estimate or under-estimate, resulting in wastage of resources or budget excess during an ongoing software project. When the software is under development, the software artefacts are in unstable states. Current estimation models are not very accurate when it comes to estimating effort based on multiple change requests. To address this issue, an estimation model is needed to aid the SPM to estimate the effort taken by the change requested to approve or reject it. In this research, a hybrid effort estimation model was developed to enable the SPM to make a more accurate decision to accept or reject a change request. The design of this model took into consideration algorithmic methods which are Software Change Impact Analysis (SCIA) and Software Change Effort Estimation (SCEE) to deal with artefacts which are either fully developed, under development or not yet being developed, hybridized with expert judgement, to provide SPM with the correct information to decide whether to accept or not a change request. The hybrid effort estimation (HEE) model was successfully developed. A prototype was built to evaluate this model. The results demonstrated improved accuracy for the hybrid model compared to the algorithmic model. Many effort estimation models exist but none works well for different types of projects. This research has demonstrated that a hybrid effort estimation model via the prototype improves estimates; taking advantage of both algorithmic and non-algorithmic model’s benefits. Experimental results showed a 17% accuracy improvement.
Description
Thesis (Doctor of Philosophy)
Keywords
Software architecture, Software engineering