Multicriteria support to sort information systems portfolio.

ABSTRACT

This article presents how to sort information system projects portfolios, based on a number of criteria used by companies according to the ELECTRE TRI method for multicriteria decision support.

The projects portfolio is set up as a step of information system planning, by using business strategies, since information systems must support companies in business strategies execution.

Thus, sorting this portfolio is the point. The ELECTRE TRI method is used as a support. This is a multicriteria method, which defines decision alternatives, in this case, IS, for pre defined categories.

Keywords: Multiple criteria analysis; Information Systems Priorities, ELECTRE TRI Method,

1. INTRODUCTION

Under the competitive market companies live in, investments in Information Systems have become essential to guarantee competitive gains. However, fast and frequent changes increase risks for bad succeeded IS investments.

It is known that changes in general usually imply on changes at business direction and this may mean changes in the most relevant IS projects and may compromise or make investment decisions feasible.

Because of that situation, the IS projects portfolio selection process and how investments are decided are so important.

Investment decisions must be steps within the IS planning process. An IS application portfolio must be designed together with companies business plans. It can be done through a planning methodology which connect business strategies and IS plans. Investments must not be decided only according to financial gains, which can not be easily measured in IS cases, but also to strategic gains. Then, the criteria for decision processes reflect the investment effectiveness.

Companies can not always see the benefits as a result from IS investments. Although they are real, these benefits are unreachable and can not be evaluated through traditional methods. Thus, some authors agree this is one of the most difficult problems to deal with in IS (Remenyi, 2000; Kumar, 1997).

As this process is usually hard, this paper presents how to sort IS projects portfolios, according to several criteria used by companies and to the ELECTRE TRI supporting method for multicriteria decisions. At the end, as an illustration, an application is shown.

2. INFORMATION SYSTEMS PLANNING

IS Strategies Plan must make part of the companies Corporate Strategies Plan. It is necessary to join the corporate and the information resources management views.

Dynamism prevents companies to achieve their strategic planning and to follow it without adaptations. Thompson e Strickland (2001) state that the companies Strategic Views represent the planned actions and the necessary reactions for unpredictable situations. In other words, companies must be pro-active when establishing their action strategies, but reactive when adapting this same strategy as responses to external or internal changes.

IS planning must come from the companies strategic view. To get a plan integration (corporate and IS planning) and to change the company strategies in IS portfolios, studying and developing methodologies and approaches that fit this planning process are really important.

Literature shows that companies have adopted different approaches (IBM, 1981; Han et al, 1998; Min et al, 1999; Kim et al 2000). Unfortunately, the application itself of the methods do not ensure a well succeeded IS planning.

Teo e Ang (2001) point out some problems which may be responsible for failures at IS planning.

* Lack of active participation and involvement of high management--this would guarantee awareness of companies strategic goals , harmony of IS plans with theses goals and the effectiveness of this plan;

* Failures at changing goals and strategies into action plans;

* Carelessness on adapting IS plans to external changes.

It is known that to guarantee effectiveness in IS plans, involvement of high management, IS experts and customers is really relevant, as well as flexibility to respond unpredictable situations and the application of a methodology which can put companies strategic views and IS applications together.

By having companies strategic view as a point to IS applications, an IS planning methodology was adopted, the BSP--Business System Planning, which was developed by IBM (IBM, 1981). BSP approach has business process as its basis to IS. Processes keep almost constant through the time, and the conditions bred by them will reflect non mobile information needs. Some aspects were simplified for this adapted version of the methodology (Almeida at al, 1992).

Firstly, the companies' strategic view is studied. Under this view, an analysis of the business processes is started. At this level, all the processes developed in the company are defined. At the next level, Engineering of Information, the IS architecture is designed. It has 3 steps: the first one is the Corporate Data, when the companies different groups of data are identified and related to the process defined at the previous level to form the Information Groups (IG). At this level, the different kinds of technology and all aspects involved are considered. According to Sprague and Watson (1989), the following IS approaches were adopted (IST): Management Information Systems (MIS), Transactional Information Systems (TIS) and Decision Support Systems (DSS). Other interesting approaches were also considered, such as the Executive Information System and Specialist Systems. The second one is the Modulation, when the two previous views are combined (IG and IST) to establish applications or even IS projects, which are called Information System Modules(ISM) and form the projects portfolio. At the last step, the type of investment will be decided by sorting this portfolio in three different categories: soon projects, not very soon ones and projects for revision.

3. INFORMATION SYSTEMS INVESTMENT DECISION

As a result of IS Planning, most companies present a large number of potential IS projects. Since their resources are limited, it is impossible to develop and introduce all projects simultaneously, so it is necessary to select them.

This paper proposes an IS application classification, to put projects together in categories which will define investments.

It is known that the criteria used for this classification are going to produce effects on the investments assertiveness, then it is important to consider companies goals and politics. There must be resources destination politics as criteria for projects portfolio classification.

Companies take different criteria for investments decision making. According to Perez (1998) 42.1% of Spanish companies have as a criterion IS needs that respond to partners IS. 92.1% of companies prefer supporting managerial decision-makings. Although management criteria are relevant, technical ones must be used. Perez (1998) states that on average 39.2% of companies consider them for their projects evaluation. Yet, in the last years, new technologies learning have not been considered for investments. Nowadays, however, 57.9% of companies have incorporated them as a criterion, Financial criteria have also been applied by 86.84% of companies for IS projects evaluation (Perez, 1998).

Since IS projects portfolio classification represents a multicriteria problem, evaluating projects and putting them in categories may not be easy because of the number of aspects or criteria to be considered.

The multicriteria decision supporting ELECTRE TRI method has been adopted as a solution and, according to a planning methodology, three criteria groups to help evaluate IS have been set up: strategic, process and technical criteria.

The strategic ones are obtained at the strategic view level, through the highest hierarchical level at companies; the process and technical ones can be defined in the following levels, when companies' processes and kinds of information technologies are designed.

These are some examples:

* Strategic Criteria: new business opportunities, high quality of products and services;

* Process Criteria: customers commitment, level criticism of process self-controlled systems;

* Technical Criteria: technological opportunities;

4. ELECTRE TRI METHOD

Literature presents four problems within multicriteria decisions issues: choice, sorting, ordering and description (Roy, 1996).

Within the most known multicriteria decision supporting methods, the outranking ones have emphasis. This outranking is binary, defined according to a group a of decision options. This decision (a) outranking (b), an S b, whether a is as good as b, according to a decision maker preferences under the available decision options for the problem (Roy, 1996).

The outranking methods have the ELECTRE (Elimination and Choice Translating algorithm) as the most representatives. They are: ELECTRE I, II, III, IV, IS and TRI.

This paper shows decision problems as a sorting issue which means putting each decision option, in IS cases, into a pre defined category. Individual evaluations of an IS defines its category, without comparisons among different IS, but among IS with defined profiles of categories limits.

The ELECTRE TRI is a multicriteria method of sorting to deal with this problem by defining alternatives of categories, according to the following steps (Mousseau & Slowinski, 1998 and YU, 1992):

* Setting up of outranking relations between alternatives and categories limits;

* Exploration of these outranking relations to define each alternative for specific categories.

Decision makers must express their preferences for each criterion through a pseudo-criterion, which means that inter-criteria information is obtained from decision makers as indifference and preference limits.

The indifference limit ([q.sub.j]([b.sub.h])) represents the biggest difference, [g.sub.j](a) - [g.sub.j]([b.sub.h]), that the decision makers accept to keep indifference between alternative a and the category limit [b.sub.h], for the criterion [g.sub.j]. The preference limit ([q.sub.j]([b.sub.h])) represents the smallest difference, [g.sub.j](a) - [g.sub.j]([b.sub.h]), which decision makers accept preferences from in favor of a to the criterion [g.sub.j].

Two kinds of inter criteria requirements affect the outranking S relation design: the importance coefficients group among criteria, which is used at the agreement test and the veto limit group, at the disagreement test.

S is designed with an index [sigma](a, [b.sub.h]) [epsilon] [0,1] which means a credibility degree of the statement a S [b.sub.h].

The alternatives definition for each category is not a direct result from relation S.

The veto limits vj([b.sub.h]) is the smallest difference gj([b.sub.h])-gj(a), incompatible with the statement a S [b.sub.h].

The partial agreement indexes cj(a,b), agreement indexes c(a,b) and partial disagreement indexes dj(a,b) are calculated as it follows:

(1) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

The statement [aSb.sub.h] is valid if [sigma](a, [b.sub.h]) [greater than or equal to] [lambda]. ([sigma]([b.sub.h] ,a) [greater than or equal to] [lambda]. [lambda] starts a cut level so that [lambda] [epsilon] [0.5,1] (Mousseau at al, 2001).

(3) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where [??] = {j [epsilon] F : dj(a, [b.sub.h]) > c(a, [b.sub.h])}

The values of [sigma](a, [b.sub.h]), [sigma]([b.sub.h], a) and [lambda] define preference situations between a and [b.sub.h]:

[sigma](a, [b.sub.h]) [greater than or equal to] [lambda] and [sigma]([b.sub.h], a) [greater than or equal to] [lambda] [right arrow] aS [b.sub.h] and [b.sub.h] Sa [right arrow] a is indifferent to [b.sub.h].

[sigma](a, [b.sub.h]) < [lambda] and [sigma]([b.sub.h], a) [greater than or equal to] [lambda] [right arrow] aS [b.sub.h] and not [b.sub.h] Sa [right arrow] a is preferible to [b.sub.h].

[sigma](a, [b.sub.h]) [greater than or equal to] [lambda] and [sigma]([b.sub.h], a) < [lambda] [right arrow] not aS [b.sub.h] and [b.sub.h] Sa [right arrow] [b.sub.h] is preferible to a.

[sigma](a, [b.sub.h]) < [lambda] and [sigma]([b.sub.h], a) < [lambda] [right arrow] not aS [b.sub.h] and not [b.sub.h] Sa [right arrow] aR [b.sub.h] and not [b.sub.h] Ra [right arrow] a is incomparable to [b.sub.h].

The procedure role of outranking relations exploration is analyzing how to compare alternatives and profiles, and defining each category these alternatives must be included in. The ELECTRE TRI method has two available procedures: the pessimist and the optimist ones (Mousseau at al, 2000).

Attn literature, several applications of ELECTRE TRI can be found in different situations, such the: Global Planning Problems, Water Resources, Sewage Treatment, Agriculture, Irrigation and Geographic IS (Lourenco and Costa, 2001; Joerin at al, 2001; Dias at al, 2002).

5. APPLICATION

Identifying impacting IS projects at business strategies and whose introduction implies on competitive advantages is the point. Thus, it is essential to define the relevant criteria for investment decisions and the IS planning methodology used may be really helpful.

After the criteria have been set up, the multicriteria method application is going to sort the ISM, under the decision makers preferences and their point of view about the relative importance of each criterion.

Evaluations on each ISM for each criterion are obtained and represented through the following matrix by considering criteria(c), IG(m) and ITS(r).

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

An analytical structure is proposed to design ISM evaluations (Almeida, 1992), which is based on extracting from decision makers the relationship intensity among IG, ITS, processes, criteria groups and put them together to obtain each ISM evaluation for each criterion.

Decision makers must establish relations between processes and strategic criteria (Figure 1.), processes and IG and then (Figure 2.), through a matrix structure (Figure 3.), the processes evaluations related to strategic criteria for IG are transferred to obtain the IG evaluation according to strategic views.

[FIGURE 1-3 OMITTED]

By comparison, these processes are evaluated under process criteria ([p.sup.c] j) and this evaluation is transferred to IG for obtaining IG evaluations according to these criteria

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Yet, experts in IS must evaluate IST related to technical criteria, [t.sup.c]i. Then, a matrix with degrees of each IG in each IST is designed:

[FIGURE 5 OMITTED]

Finally, to get ISM evaluations for each criterion, having IG and IST evaluations, participation degrees of IST in each IG are established by [m.sup.c]ij = atij.[t.sup.c]i.

The following presents an evaluation of ISM(z) for criteria(c);

[FIGURE 6 OMITTED]

The multicriteria method must be incorporated on this matrix.

Next, an application based on a case study from a previous paper is shown, by using another multicriteria method (Lyra at al, 2001).

At this study, eight criteria were used. Four strategic ones: keeping minimum level of services supply, new business opportunities, better quality patterns and staff development; two criteria of processes: customers commitment and criticism of automatic systems; and two technical criteria technological opportunities and relative technology costs.

The following processes were defined at this study: basic structure, operations, services, acquisition, technology development, human resources management, corporate support and marketing.

By relating processes and corporate data, the following Information Groups (IG) were found: Staff, Services Management and Financial Resources. The kinds of technology for dealing with information (IST) are: Transactional Information System (TIS); Management Information System (MIS); Decision Support System (DSS); Executive Information System (EIS).

A combination between IG and IST means ISM, which are the IS applications that will represent the projects portfolio to be ordered, as the following:

ISM

1 Transactional Information System of Human Resources

2 Management Information System of Human Resources

3 Decision Support System of Human Resources

4 Executive Information System of Human Resources

5 Transactional Information System of Services Management

6 Management Information System of Services Management

7 Decision support system of Services Management

8 Transactional Information System of Financial Resources

9 Management Information System of Financial Resources

10 Decision Support system of Financial Resources

11 Executive Information System of Financial Resources

Through the analytical structure previously shown, the chart 1 with ISM evaluations for each criterion can be obtained. As the decision maker can not establish priorities among projects and decide on investments, the ELECTRE TRI method is applied as it follows.

The IS were classified in 3 categories or classes:

The preference and indifference limits used were: Preference Limit - p = 0,4*g(bh)

Indifference Limit - q = 0,2*g(bh); Veto Limit - v = 0,8*g(bh); Cut Level - [lambda] = 0,5

By adopting the optimist procedure, available at the ELECTRE TRI method, the following results are obtained from the method application:

Chart 4 shows that the ISM Transactional Information System of Services Management, Management Information System of Services Management and Transactional Information System of Financial Resources were included into the projects category which will be introduced firstly; the ISM Transactional Information System of Human Resources, Management Information System of Human Resources and Executive Information System of Financial Resources were included into the projects category which will be introduced secondly. The other ISM introduction must be re-evaluated later.

6. CONCLUSIONS

This paper confirms an agreement in literature on the importance of combining companies Planning and IS Planning. The IS portfolio must come from the companies strategic views as a step of IS planning. A planning methodology which integrates corporate and IS plans and which identifies criteria is very helpful. Besides, it is clear that IS investment decisions are a multicriteria decision problem, where it is important to sort a projects portfolio and the criteria choice has an effect on the investment assertiveness. This problem can be approached by the use of a multicriteria method, as the ELECTRE TRI, worried with sorting issues.

TABLE 1--ISM EVALUATIONS

         CRITERIA

Values   0,05       0,11       0,06      0,09        0,25

ISM      Services   New        Better    Staff       Customers
         Supply     Business   Quality   Develop..   Commit.

1        0,10       0,11       0,10      0,13        0,11
2        0,10       0,11       0,10      0,13        0,11
3        0,06       0,07       0,07      0,09        0,07
4        0,03       0,04       0,03      0,04        0,04
5        0,16       0,14       0,18      0,14        0,16
6        0,16       0,14       0,18      0,14        0,16
7        0,08       0,07       0,09      0,07        0,08
8        0,12       0,12       0,09      0,10        0,11
9        0,08       0,08       0,06      0,06        0,07
10       0,04       0,04       0,03      0,03        0,04
11       0,08       0,08       0,06      0,06        0,07

Values   0,14         0,15            15
ISM      Self-        Technological   Relative
         Controlled   Opportunities   Costs
         Systems
         Criticism

1        0,10         0,05            0,04
2        0,10         0,04            0,05
3        0,06         0,07            0,08
4        0,03         0,13            0,11
5        0,16         0,07            0,06
6        0,16         0,07            0,09
7        0,08         0,07            0,08
8        0,12         0,07            0,06
9        0,08         0,07            0,09
10       0,04         0,14            0,15
11       0,08         0,23            0,20

TABLE 2--CATEGORIES

Category

3          Projects for soon introduction
2          Projects for not very soon introduction
1          Projects for evaluation

TABLE 3--CLASSES DEFINITION

    Services   New        Better    Staff      Customers   Self-
    Supply     Business   Quality   Develop.   Commit.     controled
                                                           System
                                                           Criticism
1   0,07       0,07       0,08       0,07       0,07       0,07
2   0,12       0,01       0,13       0,01       0,12       0,12

    Technological   Relative
    Opportunities   Costs

1   0,01            0,09
2   0,19            0,15

TABLE 4--ISM SORT

ISM   Category

1     2
2     2
3     1
4     1
5     3
6     3
7     1
8     3
9     1
10    1
11    2

REFERENCES

Almeida, A.T., Bohoris, G.A. STEINBERG, H., "Management information and decision support system of a telecommunication network", Journal of Decision Systems 1(2-3), 1992, 213-241.

Han, C.H., Kim, J.K., Choi, S.H. Kim, S.H., "Determination of information system development priority using quality function development", Computer ind. Engng 35(1-2), 1998, 241-244.

Dias L., Mousseau V., Figueira, J., Climaco, J., "An aggregation/disaggregation approach to obtain robust conclusions", European Journal of Operational Research 138(2), 2002, 332-348.

IBM COPORATION. "Business system planning", Information System Planning Guide. 1981.

Joerin, F., Theriault, M., Musy A., "Using GIS and outranking multicriteria analysis for land-use suitability assessment", International Journal of Geographical Information Science 15(2), 2001, 153-174.

Kim, S.H., Jang, D.H., Cho, S.H.,. "A methodology of constructing a decision path for it investment", Journal of Strategic Information Systems 9, 2000, 17-38.

Kumar, R.L., "Understanding the value of information technology enable responsiveness", The Eletronic Journal of Information Systems Evaluation 1(1), 1997, http://is.twi.tudelft.ne/ejise/

Lyra, G., Almeida, A.T., Costa, A.P.C.S. "ELECTRE multicriteria aid on priorities assignment of information systems", In Information Resources Management Association International Conference, Toronto--Canada, May 2001, p 681.

Lourenco, R.P., Costa, J.P., "Multiple criteria decision making in the new millennium". Lecture Notes in Economics and Mathematical Systems 507, 2001, 132-141.

Min, S.K., Suh, E.H., Kim, S.H., "An integrated approach toward strategic information systems planning", The Journal of Strategic Information Systems 8(4), 1999, 373-394.

Mousseau, V., Figueira, J., Naux, J.P., "Using assignment examples to infer weights for ELECTRE TRI method: some experimental results", European Journal of Operational Research 130(2), 2001, 263-275.

Mousseau, V., Slowinski, R., Zielniewicz, P., "A user-oriented implementation of the ELECTRE TRI method integrating preference elicitation support", Computer & Operations Research 27, 2000, 757-777.

Mousseau, V. Slowinski, R., "Inferring an ELECTRE TRI model from assignment examples", Journal of Global Optimization 12, 1998, 157-174.

Perez, B.E., "Information systems investment decisions in business practice: the Spanish case." European Journal of Information Systems 7, 1998, 202-209.

Remenyi, D.,. "The elusive nature of delivering benefits from it investment," The Eletronic Journal of Information Systems Evaluation. 3(1), 2000 (http://is.twi.tudelft.ne/ejise/).

Roy, B.,. Multicriteria Methodology goes Decision Aiding, Kluwer Academic Publishers. 1996

Sprague, J.R.H., Watson, H.J., Decision Support Systems--Putting Theory into Practice, Prentice-Hall. 1989.

Teo, T.S.H., Ang, J.S.K., "An examination of major is planning problems", International Journal of Information Management 21(6), 2001, 457-470.

Thompson, A.A., Strickland, A.J., Crafting and Executing Strategy: Text and Readings, Twelfth Edition, McGraw-Hill. 2001

Yu, W. ELECTRE TRI--"Aspects methodologiques et guide d'utilisation", Document du LAMSADF, 74, Universite de Paris, Dauphine, avril. 1992

A.P.C.S. Costa, Federal University of Pernambuco, Brazil

A.T. Almeida, Federal University of Pernambuco, Brazil

C.M.G. Miranda, Federal University of Pernambuco, Brazil

Author Profiles

Professor Ana Paula Cabral Seixas Costa earned her D.Sc. at the Federal University of Pernambuco, Northeast Brazil. She is an adjunct professor at the Federal University of Pernambuco.

Professor Adiel Teixeira de Almeida earned Ph.D. at the University of Birmingham, England. Currently Dr. Almeida is an adjunct professor at the Federal University of Pernambuco, Northeast Brazil.

Caroline Maria Guerra de Miranda is a doctoral student at the Federal University of Pernambuco, Northeast Brazil. She earned her M.Sc.