Back to search

Aristotle University Of Thessaloniki (AUTH)

Πρόβλεψη και εκτίμηση στην τεχνολογία λογισμικού με μεθόδους ανάλυσης δεδομένων

Abstract

dc:description

An important issue for software development industry is the delivery of high quality software on time and within budget constraints. Software organizations recognize the need for adapting effective estimation methods regarding software development aspects, such as cost and quality in order to stay competitive. In practice, exploiting software development models is not an easy issue as problems regarding the adoption and interpretation of them may arise. Classical statistical methods present limited capabilities for modeling Software Engineering data, as they do not allow the extraction of interpretable models that can deal with the inherent uncertainty of the domain. In order to solve the above problems the current thesis explores the application of improved formal estimation methods. In particular this thesis consists of three parts: 1) Comparison of methods In this dissertation a series of comparisons regarding the suitability of formal methods in estimating software cost and quality issues are performed. The methods employed are Association Rules (AR), which are applied for the first time in the software estimation domain, Decision Trees (DT), Bayesian Belief Networks (BBN), Analogy based Cost Estimation (AbE). The comparisons performed pointed out the main advantages and disadvantages of the methods. Based on the conclusions drawn by the above comparisons this thesis suggests a framework for the selection of the most suitable formal estimation method according to the estimation context. 2) Suggestions for improvement Additionally based on the results of the comparisons performed, this thesis proposes certain improvement actions regarding the application of formal methods. The combination of AR and DT is suggested for creating conceptual, explanatory estimation models. Regression via Classification framework is presented in order to automatically apply formal methods. Also the selection of parameters of AbE method is suggested using Decision Rules. Finally the application of BBNs is explored in representing various estimation problems regarding software development, as the method is able to model the uncertainty of the domain and the subjective judgment of the human estimator. 3) Practical application Finally specialized estimation models regarding the predictions of development aspects of Decision Support Systems / Knowledge Based Systems, Telecommunication applications and Open Source Software are presented. Specialized estimation models are able to consider the unique attributes of the application domain and produce estimation models that present improved accuracy.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2008

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Bibi, Stamatia
  • Μπίμπη, Σταματία

Subjects

dc:subject × 16

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/18912

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Bibi, Stamatia; Μπίμπη, Σταματία. Πρόβλεψη και εκτίμηση στην τεχνολογία λογισμικού με μεθόδους ανάλυσης δεδομένων. Aristotle University Of Thessaloniki (AUTH), 2008. http://hdl.handle.net/10442/hedi/18912