Project portfolio selection in public administration using fuzzy logic

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/236896939 Project Portfolio Selection in Public Administration Using Fuzzy Logic Article in Procedia - Social and Behavioral Sciences · March 2013 DOI: 10.1016/j.sbspro.2013.03.036 CITATIONS READS 2 255 3 authors: Lilian Noronha Nassif João Carlos Santiago Filho Public Ministry of Minas Gerais University of Minho 23 PUBLICATIONS 65 CITATIONS 5 PUBLICATIONS 2 CITATIONS SEE PROFILE José Marcos Silva Nogueira Federal University of Minas Gerais 177 PUBLICATIONS 1,824 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: SIS Project View project A Mobility Express Indicator View project All content following this page was uploaded by João Carlos Santiago Filho on 31 May 2014. The user has requested enhancement of the downloaded file. SEE PROFILE Available online at www.sciencedirect.com Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 26th IPMA World Congress, Crete, Greece, 2012 Project Portfolio Selection in Public Administration Using Fuzzy Logic Lilian Noronha Nassifa,c,*, João Carlos Santiago Filhob, José Marcos Nogueiraa a Federal University of Minas Gerais, Av. Antônio Carlos, 6627 - ICEx, Belo Horizonte, 31270-000, Brazil b Ancine, Av. Graça Aranha, 35, Rio de Janeiro, 20030-002, Brazil c Public Ministry of Minas Gerais, Av. Álvares Cabral, 1740 - 4o. andar, Belo Horizonte, 30170-001, Brazil Abstract The project selection for portfolio management in the governmental sphere is not associated with project financial return, but necessarily involves public benefits. The literature is extensive for portfolio selection when the financial approach is crucial, but little is discussed about selecting projects from the standpoint of public policies. This paper presents a method for project selection in the area of Information Technology (IT) using a tool to support decisionmaking based on fuzzy logic. A case study validated the method developed, which consists of: 1) Identification of projects; 2) Association of projects with strategic planning; 3) Categorization of projects; 4) Definition of linguistics variables and fuzzy function; 5) Definition of inference rules; 6) Function calculation; and 7) Portfolio balancing. The paper presents interesting experimental results that show the priority of projects and their success potential. The success is related with qualitative and diffuse metrics applied in a simulator for fuzzy logic. © Authors.by Published Elsevier Ltd. and/or peer-review under responsibility of IPMA © 2013 2012The Published ElsevierbyLtd. Selection Selection and/or peer-review under responsibility of IPMA Keywords: Project portfolio management; Fuzzy logic; Decision making; Information Technology. 1. Introduction A project portfolio should consider the company's strategic planning and adopt an efficient method of project selection. Because private and public organizations have specific concerns and restrictions, new methods of project selection need to be developed to address these different perspectives. The government's decision-making process is guided by a number of peculiarities such as the existence of * Corresponding author. Tel.: +55-31-3347-0918; fax: +55-31-33470918. E-mail address: lilian@dcc.ufmg.br 1877-0428 © 2013 The Authors. Published by Elsevier Ltd. Selection and/or peer-review under responsibility of IPMA doi:10.1016/j.sbspro.2013.03.036 42 Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 legal norms, interest groups, media, and all citizens. Therefore, it is very difficult to measure project success using indicators such as cost-benefit or Return of Investment (ROI). Public administrators estimate project effectiveness by using qualitative indicators. This work presents a methodology of governmental project selection. The methodology uses fuzzy logic with qualitative indicators. The methodology application provides the success potential of projects considering their prioritization. This paper is organized as follows: Section 2 presents methods of selecting projects. Section 3 highlights the main problems and goals of project selection, and describes a new methodology for governmental project selection using fuzzy logic. Section 4 shows a case study that applies the developed methods and provides results. Section 5 concludes the paper. 2. Related Works The project selection to build a portfolio is not a simple task, since resources are scarce and the demands of business exceed those limitations. Peng et al. (2005) consider project selection as a problem of allocating capital among a number of assets to maximize return on investment while minimizing risk. Mulcahy (2009) classifies the decision models of project selection into two categories: benefit measurement and constrained optimization. The benefit measurement method uses a comparative approach between projects adopting techniques such as: 1) Scoring models; 2) Economic models; and 3) Cost-benefit analysis. The constrained optimization method uses a mathematical approach, adopting techniques such as: 1) Linear Programming; 2) Integer programming; 3) Dynamic programming; and 4) Multi-objective programming. The literature presents new methods of project selection using fuzzy logic. Carlsson et al. (2007) use fuzzy logic to select projects of research and development (R & D) with the objective of avoiding inaccuracies of return. Qin et al. (2009) present mathematical models that use fuzzy logic to improve the expected value of projects. Peng et al. (2005) use credibility programming to deal with project selection problem. These articles use fuzzy logic to solve problems about return of investment. Our approach differs from the others mainly by using fuzzy logic to select projects whose qualitative benefits are more relevant than financial returns. 3. Project Selection An erroneous project selection to compose a portfolio can generate problems such as: 1) Excessive number of projects; 2) Inappropriate projects; 3) Projects disconnected from strategic objectives; 4) Unbalanced portfolio (Qin et al., 2009). In the next section, a new methodology of project selection is presented. It was elaborated to avoid these problems and to provide recommendations to people with decision-making power. 3.1. Methodology of project selection The Methodology for Governmental Project Selection using Fuzzy Logic (MGPS-fuzzy) developed in this work, extended the macro-processes of portfolio management presented in Mulcahy (2009) and included a fuzzy module. The MGPS-fuzzy is presented at Figure 1 and consists of the following processes that are executed sequentially: Identification; Association with strategic planning; Categorization; Definition of linguistic variables and fuzzy function; Definition of inference rules; Function calculation; Balancing and prioritization of portfolio. Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 Fig. 1. Methodology MGPS-fuzzy During the process "Identification", stakeholders define projects in interviews, meetings, or using forms. The project identification follows a bottom-up approach, i.e., it is elaborated by the technical team. During the process "Association with strategic planning", each project identified is associated with actions defined by the institution's strategic planning. In the process "Categorization", projects are categorized into groups of same similarities. Processes "Definition of linguistic variables and fuzzy function", "Definition of inference rules", and "Function calculation" are part of the fuzzy module, which will be detailed in section 3.2. 3.2. Fuzzy module The fuzzy logic (Zadeh, 1965) is an extension of boolean logic that introduces the notion of sets with partial membership. It is used in many fields where systems must deal with imprecision and uncertainty. With the use of linguistic variables and rules of inference, fuzzy logic provides a mathematical framework capable of representing not only the knowledge of experts as well as the preferences of decision makers in investment projects (Wang & Hwang, 2005). The fuzzy module in the MGPS-fuzzy consists of three processes that interact strongly: 1) Definition of linguistic variables and fuzzy function; 2) Definition of inference rules; and 3) Function calculation. In the process “Definition of linguistic variables and fuzzy function”, the values 0 through 10 are associated to each variable to represent a benefit or an importance of a project. The fuzzy function represents the result of linguistic variables correlation. The linguistic variables and the fuzzy function must be defined together with stakeholders. In the process “Definition of inference rules”, it is necessary to define rules that correlate linguistic variables and fuzzy function to obtain interpretations as follows: • the more variable-1 decreases and variable-2 decreases, the better is the result; • the more variable-2 decreases and variable-3 increases, the better is the result; • the more variable-1 increases and variable-2 increases, the better is the result; According to the rules described above, variable-1, variable-2, and variable-3 correspond to linguistic variables, which are the inputs to the fuzzy model; and result corresponds to the fuzzy function, which is the output of the fuzzy model. In the process “Function calculation”, the inference rules and linguistic variable values are used for each project recognized at the “Identification” process. 43 44 Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 4. Case Study Using MGPS-fuzzy The methodology MGPS-fuzzy presented in section 3 was applied to a case study in a government company with 500 employees and about 40 IT professionals. The IT area is traditionally composed by subareas of infrastructure and system development. This work focuses on project selection in the subarea of infrastructure. In the following, the methodology is applied. 4.1. Project identification The identification of IT projects was conducted in three meetings with the infrastructure team and 25 projects were identified as described in Table 1. Table 1. Identified projects Project ID Project name 1 Data security policy 2 Definition of a security and risk management team 3 Definition of a treatment and incident response network computing team 4 Definition of process to start services in production environment 5 Demand management 6 Desktop backup 7 Desktop remote management 8 Distribution of new workstations 9 Infrastructure management 10 Internet access policy 11 Internet link upgrade 12 Link upgrade (MPLS) 13 Migration from 32bits to 64bits servers 14 Password security policy 15 Review of Active Directory 16 Review of CPD environment 17 Review of e-mail 18 Review of firewall 19 Review of network structure 20 Review of service desk 21 Review of Squid 22 Server backup review 23 VoIP 24 VPN establishment for internal access and software factory 25 Wireless network Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 4.2. Association of projects with strategic planning In our case study, the interfacing of IT projects with other areas of the enterprise occurs as shown in Figure 2. The client area represents all areas in the enterprise where the projects are strongly linked to strategic planning. The client area frequently demands projects to IT area. Most projects require the development of systems. The system development team requires a subset of projects in the infrastructure to support the demands of communication and storage. The area of infrastructure, in turn, needs to keep and improve the current operational environment. Therefore, some of infrastructure projects related to the strategic planning are mediated by the requirements identified by the system development team. The strategic planning of the case study has generic actions defined to IT area, namely: organize the systems and IT services; build and maintain a reliable public database; organize and disseminate information. Fig. 2. Association of projects with strategic planning 4.3. Categorization of projects The categorization of projects will assist the processes of defining linguistic variables and portfolio balancing. In this study, we identified categories according to actions defined in the strategic planning and demands from the system development team, as shown in Figure 2. Table 2 shows the grouping of projects in categories of affinity. The categories represent specific actions to address actions defined to IT area in the strategic planning. Table 2. Categories identification according to strategic planning Category Project ID Availability of the environment 9, 16, 22 Expansion 8, 11, 12 Improved customer service 6, 7, 17, 20, 25 Innovation 23 New systems (support activities of the development team) 4, 5, 13, 24 Requirement of the laws 2, 3 Security 1, 10, 14, 15, 18, 19, 21 4.4. Definition of linguistic variables and fuzzy function The linguistic variables were defined by stakeholders based on the benefits or difficulties in project execution. The variables chosen were: cost (given the necessary public investment and annual budget constraint in the IT area); external dependence (assuming that an IT project that interacts with many areas 45 46 Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 of the company has a greater risk of not succeed); and visibility (assuming that the project impact on users can cause a good impression of the administration). Therefore, in the case study, the cost, the external dependence, and the visibility correspond to linguistic variables; and the project success potential corresponds to the fuzzy function. 4.5. Definition of inference rules Stakeholders defined the inference rules, establishing a correlation between linguistic variables and fuzzy function, as follows: • the more cost decreases and external dependence decreases, the better is the project success potential; • the more external dependence decreases and visibility increases, the better is the project success potential; • the more cost decreases and visibility increases, the better is project success potential. Where each variable has the following values and interpretations: External dependence: 0 (good), 10 (bad); Cost: 0 (good), 10 (bad); Visibility: 0 (bad), 10 (good). From these correlations, we constructed 27 inference rules. Table 3 depicts examples of inference rules that will be described in the fuzzy model. Table 3. Inference rules depicted in the fuzzy model Cost (input) Visibility (input) Dependence (input) Project success potential (output) High Low High Weak Low High Low Strong Low High Medium Strong High High High Medium High Low High Weak Low Medium Medium Medium Two important concepts must be considered when inserting or extracting data from the fuzzy model: fuzzification and defuzzification (Shaw, 1998). The fuzzification process interprets the scale of numerical values of linguistic variables to a description apparently informal (Low, Medium, High). This translation is significant because the fuzzy logic helps decision makers where the uncertainty is strongly present. In the case study, stakeholders defined values for each project presented in Table 1, on a scale of 0 to 10, for each linguistic variable. Table 4 shows the fuzzification of values to informal values Low, Medium, and High; where Low represents the values 0 to 3; Medium represents the values 4 to 6; and High represents the values 7 to 10. The process of defuzzification is the reverse of the process of fuzzification, i.e. it transforms subjective data into numbers. This process will be used in the function calculation process. Table 4. Fuzzification of values for the linguistic variables associated with each project Informal Values External dependence Cost Visibility Low 12, 11, 22, 9, 3, 2, 7, 20, 15, 18, 19, 21 12, 11, 3, 2, 17, 13, 5, 4, 24, 15 12, 16, 22, 3, 2, 25, 13, 5, 4, 24, 10, 14, 1, 18, 19, 21 Medium 8, 25, 6, 14, 5, 13, 17, 24 9, 25, 20, 18, 19, 21 8, 9, 7, 20, 15 High 23, 16, 10, 1 23, 2, 16, 22, 7, 6, 10, 14, 1 23, 11, 6, 17 Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 4.6. Calculation of potential The success potential of each project was calculated using the tool called Fuzzy Logic Toolbox, available in Matlab (MathWorks, 2012). The simulation model is formed by rules defined in Table 3 and linguistic variables of each project, defined in Table 4. After these calculations, the results were defuzzyficated, i.e., the terms Weak, Medium, and Strong associated to success potential of each project were transformed into numerical values. The graphs shown in Figures 3, 4, and 5 represent functions defuzzyficated, which correlate each pair of linguistic variables. Figure 3 shows that project success potential increases when external dependence decreases and visibility increases. Figure 4 shows that project success potential increases when external dependence decreases, and cost decreases. Figure 5 shows that project success potential increases when cost decreases, and visibility increases. The calculation of the success potential for each project is shown in Table 5. Fig. 3. Visibility versus external dependence Fig. 4. Cost versus external dependence 47 48 Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 Fig. 5. Visibility versus cost 4.7. Portfolio balancing In the case study, the portfolio balancing was done by the project prioritization given by different stakeholders. They prioritized projects based on their evaluation of subjective variables, such as availability of human resources and project completion prediction. The list of projects was revised using four priority classes given by three different stakeholders: the infrastructure team, the systems development team, and the manager of user services. Table 5 presents the final results, where the priority of each project was calculated from the average of different views with weights 3, 2, and 1 respectively to the areas of infrastructure, systems development, and manager of user services. Projects are ranked in decreasing order of priority followed by the project success potential. 5. Conclusions The project selection to compose a portfolio is not a trivial task. Several variables must be considered, such as the dependence of external areas, the project cost, and the project visibility. In the case study conducted, initially, the IT infrastructure team identified a list of necessary projects. After this survey, projects were categorized according to affinity groups and associated with actions of the company strategic planning. Subsequently, stakeholders defined the linguistic variables, the fuzzy function, and the inference rules for the model. Finally, using the previous information, a simulator of fuzzy logic calculated the success potential for each project. The methodology developed is a new approach of selecting projects to compose a portfolio using a technique of artificial intelligence, fuzzy logic. The model can help in the decision-making process where uncertainty is strongly present. Our approach considers qualitative metrics because it is applied in the governmental sphere, which must select projects from the point of view of benefits instead of the financial return perspective. 49 Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 Table 5. Project portfolio ID Project name Project success potential Priority 11 Internet link upgrade 0.863 1 3 Distribution of new workstations 0.5 1 22 Server backup review 0.5 1 9 Infrastructure management 0.5 1 1 Data security policy 0.5 1 19 Review of network structure 0.5 1 16 Review of CPD environment 0.137 1 14 Password security policy 0.863 2 10 Internet access policy 0.646 2 3 Definition of treatment and incident response network computing team 0.549 2 2 Definition of security and risk management team 0.549 2 17 Review of e-mail 0.549 2 12 Link upgrade (MPLS) 0.5 2 25 Wireless network 0.5 2 24 VPN establishment for internal access and software factory 0.5 2 15 Review of Active Directory 0.5 2 6 Desktop backup 0.646 3 7 Desktop remote management 0.5 3 20 Review of service desk 0.5 3 13 Migration from 32bits to 64bits servers 0.5 3 5 Demand management 0.5 3 4 Definition of process to start services in production environment 0.5 3 18 Review of firewall 0.5 3 21 Review of Squid 0.5 3 23 VoIP 0.5 4 References Carlsson, C., Fullér, R., Heikkilä, M., & Majlender, P. (2007). A fuzzy approach to R&D project portfolio selection. International Journal of Approximate Reasoning, 44(2), pp. 93-105. Mulcahy. R. (2009). PMP Exam Prep, Sixth Edition: Rita's Course in a Book for Passing the PMP Exam. (6th ed.). RMC Publications, Inc. Peng, J., Mok, H. M. K., Tse, & W-M.(2005). Credibility Programming Approach to Fuzzy Portfolio Selection Problems. Proceedings of the Fourth International Conference on Machine Learning and Cybernetic (pp. 2523-2528). Guangzhou. 50 Lilian Noronha Nassif et al. / Procedia - Social and Behavioral Sciences 74 (2013) 41 – 50 Qin, Z., Li, Z., & Ji, X. (2009). Portfolio selection based on fuzzy cross-entropy. Journal of Computational and Applied Mathematics, 228(1), pp. 139-149. Shaw, I. S. (1998). Fuzzy Control of Industrial Systems: Theory and Applications. Kluwer Academic Publisher. Massachusetts. MathWorks (2010). Matlab 6.5. The MathWorks Inc.. Natick, Massachusetts. Wang, J., & Hwang, W.-L. (2007). A fuzzy set approach for R&D portfolio selection using a real options valuation model. Omega, 35, pp. 247-257. Zadeh, L. A. (1965). Fuzzy sets. Information and control, 8(3), pp. 338-353. View publication stats