An Investigation into Software Estimation Methods

Hamdan, Khaled (2009) An Investigation into Software Estimation Methods. Doctoral thesis, University of Sunderland.

[img]
Preview
PDF
Khaled_Hamdan.pdf - Accepted Version

Download (3MB)

Abstract

There are currently no fully validated estimation approaches that can accurately predict
the effort needed for developing a software system (Kitchenham, et al, 1995).
Information gathered at the early stages of system development is not enough to
provide precise effort estimates, even though similar software systems may have been
developed in the past. Where similar systems have been developed, there are often
inherent differences in the features of these systems and in the development process
used. These differences are often sufficient to significantly reduce estimation accuracy.
Historically, cost estimation focuses on project effort and duration. There are many
estimation techniques, but none is consistently ‘best’ (Shepperd, 2003).
Software project management has become a crucial field of research due to the
increasing role of software in today’s world. Improving the functions of project
management is a main concern in software development organisation. The purpose of
this thesis is to develop a new model which incorporates cultural and leadership factors
in the cost estimation model, and is based on Case-Based Reasoning. The thesis
defines a new knowledge representation “ontology” to provide a common understanding
of project parameters. The associated system uses a statistically simulated bootstrap
method, which helps in tuning the analogy approach before application to real projects.
This research also introduces a new application of Profile Theory, which takes a formal
approach to the measurement of leadership capabilities.
A pilot study was performed in order to understand the approaches used for cost
estimation in the Gulf region. Based on this initial study, a questionnaire was further
refined and tested. Consequently, further surveys were conducted in the United Arab
Emirates. It was noticed that most of the software development projects failed in terms
of cost estimate. This was due to the lack of a precise software estimation model.
These studies also highlighted the importance of leadership and culture in software cost
estimation.
Effort was estimated using regression and analogy. The Bootstrap method was used to
refine the estimate of effort based on analogy, with correction for bias. Due to the very
different nature of the core and support systems, a separate model was developed for
each of them. As a result of the study, a new model for identifying and analysing was
developed. The model was then evaluated, and conclusions were drawn. These show
the importance of the model and the factors of organisational culture and leadership in
software project development and in cost estimation. Potential areas for future research
were identified.

Item Type: Thesis (Doctoral)
Subjects: Computing > Software Engineering
Divisions: Collections > Theses
Depositing User: Barry Hall
Date Deposited: 15 Feb 2013 16:22
Last Modified: 07 Mar 2017 19:41
URI: http://sure.sunderland.ac.uk/id/eprint/3557

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year