**Framework for Optimal Selection Using Meta‐Heuristic Approach and AHP Algorithm**

Ramo Šendelj and Ivana Ognjanović

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/63991

#### **Abstract**

[22] Matyas, D., Pelling, M. 2012. *Disaster Vulnerability and Resilience: Theory, Modeling and Prospective*. Report produced for the Government Office of Science, Foresight project

[23] Mayunga, J. S. *Understanding and Applying the Concept of Community Disaster Resilience: A capital‐based approach*. A Draft Working Paper Prepared for the Summer Academy for Social Vulnerability and Resilience Building, 22–28 July 2007, Munich, Germany.

[24] Miller, F., H. Osbahr, E. Boyd, F. Thomalla, S. Bharwani, G. Ziervogel, B. Walker, J. Birkmann, S. Van der Leeuw, J. Rockström, J. Hinkel, T. Downing, C. Folke, and D. Nelson. 2010. Resilience and vulnerability: complementary or conflicting concepts? *Ecology and Society* 15(3):11. [online] URL: http://www.ecologyandsociety.org/vol15/

[25] Olcina, J. 2008. *[Changes in the territorial, conceptual and method of natural hazards consid‐ erations]*. In X International Colloquium Geocrítica, Ten years of changes in the world,

[26] Saaty, T., Vargas, L. 1994. *Decision Making in Economic, Political, Social and Technological Environments. With the Analytic Hierarchy Process*. RWS Publications, USA. 325 p.

[28] Saaty, T., Peniwati, K. 2008. *Group Decision Making. Drawing Out and Reconciling*

[29] Siddayao, Generino P., Sony E. Valdez, and Proceso L. Fernandez. 2014. Analytic Hierarchy Process (AHP) in Spatial Modeling for Floodplain Risk Assessment. *International Journal of Machine Learning and Computing* 4(5):450–457. doi:10.7763/IJMLC.

[30] Wachinger, G., Renn, O. 2010. Risk Perception and Natural Hazards. CapHaz‐Net WP3 Report, DIALOGIK Non‐Profit Institute for Communication and Cooperative Re‐ search, Stuttgart. [online] URL: http://caphaz‐net.org/outcomes‐results/CapHaz‐

[31] Whitaker, R. Validation examples of the AHP and ANP. 2007. *Mathematical and*

[32] Yamin, L. E., F. Ghesquiere, O. D. Cardona, and M. G. Ordaz. 2013. *[Probabilistic modeling for disaster risk management: the case of Bogotá, Colombia]*. Banco Mundial, Universidad de los Andes. [World Bank, Universityof the Andes] 188pp. 2013. [In

in geography and social sciences, 1999–2008. Barcelona. 2008.[In Spanish].

[27] Saaty, T. 1996. *Decision Making for Leaders*. RWS Publications, USA. 315 p.

*Differences.* RWS Publications, USA. 385 p.

Net\_WP3\_Risk‐Perception.pdf

*Computer Modeling* 46(7–8):840–859.

'Reducing Risks of Future Disasters: Priorities for decision makers'. 72 pp.

192 Applications and Theory of Analytic Hierarchy Process - Decision Making for Strategic Decisions

Retrieved October 2011.

iss3/art11/

2014.V4.453.

Spanish].

Many real‐life decisions are focused on selecting the most preferable combination of available options, by satisfying different kinds of preferences and internal or external constraintsandrequirements.Focusingonthewell‐knownanalyticalhierarchicalprocess (AHP) method and its extension CS‐AHP for capturing different kinds of preferences over two‐layered structure (including conditionally defined preferences and preferen‐ ces about dominant importance), we propose a two‐layered framework for identifying stakeholders' decision criteria requirements and employ meta‐heuristic algorithms (i.e., genetic algorithms) to optimally make a selection over available options. The proposed formal two‐layered framework, called *OptSelectionAHP*, provides the means for optimal selection based on specified different kinds of preferences. The framework has simulta‐ neously proven optimality applied in software engineering domain, for optimal configuration of business process families where stakeholders' preferences are defined over quality characteristics of available services (i.e., QoS attributes). Furthermore, this domainof applicationis characterizedwithuncertaintyandvariabilityinselectionspace, which is proven and does not significantly violate the optimality of the proposed framework.

**Keywords:** AHP, CS‐AHP, genetic algorithms, optimal selection, two‐layered criteria structure, user preferences

### **1. Introduction**

Many real‐life decisions are made by considering and analyzing different kinds of preferen‐ ces with different impacts on final selection of option among available, with more often optimization problems defined in terms of both hard and soft constraints. The modeling of user

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

preferences is a great challenge, as it is difficult to express human opinion in a way that can be easily processed by computers [1]. Researchers in many different fields (such as, economics, riskmanagement,decisiontheory,socialchoicetheory,operationalresearch,intelligentsystems, databases, etc.) studied the representation of preferences, its processing and practical use [2].

Additionally, there is one more demand on allowing automated support for selection of the most preferable combination of available options, with respect to specified preferences and constraints if exist. In order to develop a framework for both, representing and reasoning about different kinds of requirements, the following two issues should be carefully analyzed: (i) comprehensive model for presentation of different kinds of preferences, on which bases develop (ii) the approach for automated selection by optimizing the fitting degree of satisfying specified preferences.

Traditional elicitation methods are typically developed based on pair‐wise comparisons, priority groups, networks for decision‐making and cumulative ratings [3, 4]. They usually collect independent preferences, under the mutual preference independence (MPI) hypothesis [5], which means that a user's preference for an option is independent of the other options [1]. However, the MPI hypothesis is not always true in practice [6] since people often express conditional preferences as to be more natural to the human way of thinking [1]. This is why well‐accepted and widely used analytical hierarchical process (AHP) algorithm proposed by Saaty [7] has been recently extended in order to handle conditionally defined preferences over two‐layered hierarchical structure, namely CS‐AHP [2].

On the other hand, there is a wide range of different optimization and search techniques that have been used for solving the optimization problems [8]. Classical techniques [such as linear programming (LP)] are often distinguished as straightforward deterministic algorithms which are distinct from meta‐heuristic search, such as, hill climbing [9], simulated annealing [10] and genetic algorithms (GAs) [11]. However, these deterministic optimization algorithms are often inapplicable in many real‐world problems, because the problems have objectives that cannot be characterized by a set of linear equations [12].

In this chapter, we demonstrate how the selection processes can be automated in a more scalable manner by using AHP (and its extension for handling conditionally defined prefer‐ ences, CS‐AHP) and Genetic Algorithms (GA) for presentation and analyses of different kinds of preferences and solving the problem of the optimal selection over the set of available options. Our framework, called *OptSelectionAHP*, provides the following major benefits to the process of prioritization, decision making and optimal selection:


In the rest of the chapter, we first introduce an illustrative example about real‐life decision‐ making scenarios that will be employed throughout the chapter (Section 2). In Section 3, we introduce the whole approach and formalize selection of problem over available options and different kinds of preferences defined over two‐layered decision criteria structure (as defined by CS‐AHP algorithm). Section 4 formalizes quality measurements induced by CS‐AHP outputs, on which bases optimal selection goals are formalized and genetic algorithms adopted for their solving. Finally, Section 5 presents simulation analyses in the area of business process families for the problem of optimal service configuration, but it is clear that our proposed work in this chapter is general enough to be applied to any optimal selection processes and domains. The critical review of methods and frameworks from related work is presented in Section 6 before the chapter is concluded.
