The Department of Computer Science at the University of Cyprus cordially invites you to the PhD Defense entitled:

Exploring Software Cost Modelling and Estimation with Computational Intelligence

Speaker: Efi Papatheocharous
Affiliation: University of Cyprus, Cyprus
Category: PhD Defense
Location: Room 148, Faculty of Pure and Applied Sciences (FST-01), 1 University Avenue, 2109 Nicosia, Cyprus (directions)
Date: Monday, December 19, 2011
Time: 10:30-11:30 EET
Host: George Papadopoulos (george AT and Andreas Andreou (

Software cost estimation (SCE) is the art of balancing time and resources to optimally budget a project. An essential requirement is to estimate the schedule, cost and human effort required to complete the project with adequate accuracy before the project commences. Estimates performed at an acceptable (early) point in time during the project life-cycle may be also considered useful to the project. SCE models proposed within this area of research consist of mathematical algorithms or parametric relations used to approximate the most dominant cost for developing a software product, the human effort, in terms of person-months. Thus SCE is considered one of the basic project management processes to support efficiently the activity of resource allocation. Although numerous SCE models and techniques have been proposed by researchers, recent research reports 60-80% of projects overrun software cost estimates by 30-40%. The present dissertation proposes models and techniques, based on the factors of People – Process – Product, for improving software cost estimation accuracy and increasing comprehensibility over the risks occurring. Moreover, the essential factors that affect productivity are identified and explored. Particularly, this diatribe adopts two approaches: A quantitative and a qualitative. The quantitative approach, aims at improving SCE accuracy, reliability and generalisability, by exploring Computational Intelligent (CI) models and techniques, such as Artificial Neural Networks (ANN), Evolutionary Algorithms (EA), Fuzzy Logic (FL), and hybrid forms of the aforementioned techniques. Moreover, the main target is to develop SCE models of a practical value, i.e., dealing with the inherent uncertainty of the software engineering data and producing relatively ‘early’ (i.e., post specifications) estimations. The qualitative approach extends the numerical and empirical CI investigations, by employing Fuzzy Cognitive Maps (FCM) and Influence Diagrams (ID), which facilitate the exploration of the relationships between qualitative (‘soft’) cost factors and effort. It also visually reveals the contribution of attributes in SCE and enhances our understanding of the cause-and-effect dependencies. The results and observations obtained from the experimental studies of this dissertation reveal that considerable benefits are gained by CI-based methods employed in cost prediction improvement and in understanding which factors are more significant or influential in the process of SCE. A variety of Feature Subset Selection (FSS) methods are applied that facilitate in identifying and excluding the less influential cost factors from the models, which, in turn, lower the overall time and effort required to measure, collect and maintain these factors. The SCE models proposed in this thesis are proven viable, practical alternatives through extensive experimentation with widely known and used benchmark data obtained from the relevant literature.

Short Bio:
Efi Papatheocharous is a PhD Candidate and Research Associate at the Department of Computer Science at the University of Cyprus, where she carries out research and studies primarily in the area of Software Engineering with specific research interests in Computational Intelligence and Software Cost Estimation. After graduating from the University of Manchester with an MSc degree in Advanced Computer Science with Information Communication Technologies Management from the Computer Science Department, in December 2005, and after attaining an undergraduate BSc in Computer Science from the Computer Science Department at the University of Cyprus, in June 2004, she has been collaborating in a number of research projects at a National and European level and has co-authored several papers presented in International Conferences, published in peer reviewed books and scientific journals.

  Other Presentations Web:
  Colloquia Web: