Decision making approaches for software quality function deployment

ZEYNEP ŞENER

Decision making approaches for software quality function deployment

Thesis Type: Doctorate

Institution Of The Thesis: Galatasaray Üniversitesi, Mühendislik ve Teknoloji Fakültesi, Endüstri Mühendisliği, Turkey

Approval Date: 2010

Thesis Language: English

Student: ZEYNEP ŞENER

Supervisor: EYÜP ERTUĞRUL KARSAK

Open Archive Collection: AVESIS Open Access Collection

Abstract:

With the rapid development of software industry, improving software quality has gained increasing importance. Software quality function deployment (SQFD), as a technique for improving the quality of software development process to create products responsive to customer expectations, is used to maximize customer satisfaction. As a front-end technique, SQFD is an adaptation of the house of quality (HOQ), the most commonly used matrix in the traditional quality function deployment (QFD) methodology. The objective of the HOQ is to determine the target levels of engineering characteristics of a product to maximize customer satisfaction. In this work, fuzzy regression is used to estimate the parameters of functional relationships between customer needs and engineering characteristics and among engineering characteristics. The inherent fuzziness of relationships in SQFD modelling justifies the use of fuzzy regression. Then, decision making approaches based on fuzzy regression and optimization are proposed to determine the target levels of engineering characteristics. The crisp model which considers only the center value estimates from fuzzy regression is extended to fuzzy case by taking into account spread values of fuzzy parameters. Furthermore, the fuzzy multiple objective decision making (MODM) framework presented in this work enables to include other enterprise related objectives in the decision process. Finally, the fuzzy MODM approach is extended by considering future requirements to determine target levels of the new or improved software that meet customer needs at the time when the product reaches the market. A search engine quality improvement problem is presented to illustrate the application of the proposed approaches.