Back to search

Aristotle University Of Thessaloniki (AUTH)

Στατιστικά μοντέλα ανάλυσης και εκτίμησης δεδομένων διοίκησης λογισμικού

Abstract

dc:description

Estimating the cost or the duration of a software project is an important issue in software engineering. This estimation is usually made with statistical models which are based on historical databases. Since most of the software cost datasets include missing values, the problem of their completion arises. The present dissertation deals with the estimation of the cost or the duration and also with the handling of missing data, using statistical methods and models which are built on software cost data. More specifically, the contribution of the dissertation is focused on the following fields: First of all the ordinal regression (OR) is presented for the estimation of the cost of software projects, in order to produce categorical cost estimations according to estimated probabilities. The method also provides the capability for point estimation, in order to be able to be compared with other traditional methods. Then, the method of multinomial logistic regression (MLR) is applied for the estimation of missing values in categorical data. The method is first applied on a complete data, using artificially generated missing values, and then on a dataset with real missing data. Finally, a statistical framework is presented, in order to study the distribution of the duration of software projects and also the most important factors that affect it. The statistical methodology of survival analysis (SA) is proposed, having as basic feature the construction of probabilistic models from software data with not only completed projects but also from projects which are still active (ongoing projects). In summary, the above methods are able to contribute positively in dealing with the problems which each one is involved.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2007

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Sentas, Panagiotis
  • Σέντας, Παναγιώτης

Subjects

dc:subject × 16

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/20355

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

Sentas, Panagiotis; Σέντας, Παναγιώτης. Στατιστικά μοντέλα ανάλυσης και εκτίμησης δεδομένων διοίκησης λογισμικού. Aristotle University Of Thessaloniki (AUTH), 2007. http://hdl.handle.net/10442/hedi/20355