Multi-Criteria Decision Support System

Multi-Criteria Decision Support System

Distributed Environment” as an area of future research, tried to provide a framework of developing a new decision support tool for general people. The new decision support tool was developed combining Multi Attribute Utility Theory (MATT) and Hypothetical Equivalents and Nonequivalent Method (HEMI). The new tool is designed to be easily understandable, easy to administer, taking less time and efficient. A step by step explanation of the new tool with the help of an example is presented in the paper. At the end, different ways to refine the tool and discussion on how to build the decision support system is presented.

This type of decision support tool could be adopted by online retail sellers to provide their users a way of efficiently comparing between different alternatives. Keywords Multi-criteria Decision Making, MACE, Multi Attribute Utility Theory, MATT, Hypothetical Equivalents and Nonequivalent Method, HEMI, Decision Supporting in a Distributed Environment. Raman: Decision Support System in a Distributed Environment 1. Introduction Multiple Criteria Decision Making (MACE) can be defined as the study of methods and procedures by which concerns about multiple conflicting criteria can be formally incorporated into the management planning process [1].

MACE evaluates the advantages and disadvantages of alternatives based on multiple criteria and produces a ranking of alternatives [2]. Usually, there does not exist a unique optimal solution for such problems and it is necessary to use decision maker’s preferences to differentiate between solutions [3]. There have been important advances in the field of MACE since the start of the modern multiple cartels calicles making Oligopolies In ten early have traditionally sought to support corporate managers. There has not been enough search to provide decision support systems for the general people.

Wallabies et al. [4] in their paper has identified this as an area of future research and wrote: “Decision supporting a distributed environment is somewhat different from what we had expected in the early sass. For example, what is a user? Our fields have traditionally sought to support corporate managers. However, household consumers need support for purchasing decisions. What kind of decision support do they want in an Internet or mobile environment? The problem may not be one of having insufficient information, but rather one of having “too much” or an unknown quality f information.

We may have to filter information. This is a potential application and development area for MACE/MATT. ” This paper, considering “Decision Supporting in a Distributed Environment” as an area of future research, tried to provide a framework of developing a decision support tool for general people. This type of decision support tool could be adopted by online retail sellers to provide their users a way of efficiently comparing between different alternatives. However, decision support system for general mass would be different from traditional MACE.

As it involves general people, there are a number of challenges to overcome: 0 0 0 0 The tool must be easy to administer. It must take less time. The contents/questions must be understandable to the general people. It must be as efficient as other validated tools. Raman: Decision Support System in a Distributed Environment There are several effective tools for multi-criteria decision making; Multi Attribute Utility Theory (MATT) is one them. Apart from the efficiency of MATT, it is a difficult tool to administer and it is time consuming too. Similar disadvantages apply for other methods too.

Therefore, to overcome the challenges mentioned above, this paper tried to combine wow methods. In the combined method, utility functions for each attribute would be calculated using MAIMS approach and then relative weights to different criteria would be assigned using Hypothetical Equivalents and Nonequivalent Method (HEMI). 2. Literature Review Unaided human decision making often exhibits inconsistencies, irrationality and suboptimal choices [5]. Decision theory was originally developed because people are often dissatisfied with the choices they make, and find it difficult to determine which choice best reflects their true preferences [6].

The goal of decision theory is not to redirect or mimic the choices of human decision maker, rather to help human make better decisions [6]. Multi Attribute Utility Theory [6], developed by Kenney and Raffia, attempts to maximize a decision maker’s utility or value (preference) which is represented by a function that maps an object measured on an absolute scale into the decision maker’s utility or value relation [2]. MAIMS involves a single decision maker who participates and answers a number of lottery questions and based on his answers TTY Tunnels Tort Deterrent attribute Ana attribute-scaling parameters are generated.

MATT has been an efficient tool in the field of multiple criteria decision making and was used effectively to solve practical problems [7] [8]. The HEMI approach determines the attribute weights using a set of preferences rather than selecting weights arbitrarily based on intuition or experience [9]. The approach is based on developing a set of hypothetical alternatives and then asking the decision maker to rank them to have several inequalities. These inequalities are then used to analytically solve for the theoretically correct set of attribute weights.

Combining MAIMS with HEMI would have the advantages of both methods while educing some of the unfeasible characteristics in the context of distributed environment. The combined method considers that the utility functions can be non- linear. The use of HEMI would also require less time to produce attribute weights than MATT. 3. Step by Step Explanation of the New Method with an Example To explain the new method, a problem of “buying a used car” is considered. In this problem, there are five alternatives and three criteria.

Table 1 gives different criteria value for different alternatives. Table 1: Different Criteria Value for Alternatives. Price ($) Toyota Ionians Ford Mercury Honda 2450 3000 2800 2700 2800 Age 178 155 162 211 190 Mileage 21 26242628 The decision maker wanted to minimize price and age (in thousands of miles driven) of the car while maximizing mileage. The problem can be solved through following steps: Step 1: Generate Utility Function for Each Attribute Utility Functions for each attribute are to be developed assuming an exponential distribution [7] [8].

Exponential estimation of utility functions and procedures of asking questions are presented in several papers [7] [8]. To estimate different parameters at first the best and worst payoffs for all criteria were identified. Then the decision maker was asked to estimate the value of certainty equivalent (CE) for all the criteria. Using the value of CE, risk tolerance (ART) for all criteria was calculated. The parameters “A” and “B” were then calculated for all criteria. All the calculated values are presented in Table 2. Table 2: Best,worst payoffs price (S) Best payoff worst payoff CE ART A B 2450 3000 2600 534 -0. 6-152. 88 Age 155 211 170 34. 12-0. 24 -116. 51 Mileage 2821 24 703. 95 101. 06 104. 13 Raman: Decision Support System in a Distributed Environment Step 2: Calculate Utilities for All Criteria Value for Each Alternative Using the utility function developed in the first step, utilities for all criteria value for each alternative were calculated and presented as matrix U. Step 3: Estimate Attribute Weight Using HEMI For the hypothetical equivalents, a 33-1 fractional factorial design presented in [9] was considered. The hypothetical alternatives are given in Table 3.

Table 3: Hypothetical Alternatives Alternatives A B C DE F G H I pence (S) 0 1225 24500 1225 24500 1225 2450 Age 0 77. 5 155 77. 5 155 0 155 0 77. 5 Mileage 028 14 1402828 140 Decision Maker (DMS) was asked to rank three alternatives at a time, between the sub groups {A, B, C}, {D, E, F}, and {G, H, l}. The ranks were as follows 0 0 0 Using the ranking done by the DMS six inequalities were obtained, (1) (2) Raman: Decision Support System in a Distributed Environment (3) (4) (5) (6) Now, using equation (1) through (6) following optimization problem was formed. Attributes. ( ) ‘s are the weights for different Solving the optimization problem, produce the following weights presented as a W matrix: [ Step 4: Ranking the Alternatives Multiplying matrix U with matrix W reduced the overall utilities for all the alternatives. The alternative that has highest overall utility was ranked as 1 and that was denoted as the best of all the alternatives. In case of the example, Toyota car model is the best option. Overall Utility Toyota Ionians Ford Mercury Honda Rank ] 0. 5979 0. 4288 0. 4261 0. 3881 0. 4272 12453 4.

Discussion and Future Research The tool that is explained in this paper is not validated. Therefore, experiments need to be designed to validate the new tool. The validation process would also find the scope of the decision making problems that the tool is capable of solving. During validation of the tool, a study on how well the tool produces preferences that match the user mental preferences can also be done. This study would help to make necessary changes in the tool to incorporate the effect of human emotion during decisions making.

The tool was developed to be easily understandable, easy to administer, taking less time Ana to De inclement . Silence ten unrestrainedly Is Important Tort ten tool to De effective, several improvement opportunities could be considered. The way of asking lottery questions to develop the utility functions can be refined to make it easier. Claudio et al. [10] described a new way of asking lottery questions to the decision maker. Again, decision maker’s utility or value may not depend on the levels of performance on different criteria, but instead on whether the levels meet a target or threshold on one or more criteria [4] [11].

For example, in case of buying an air ticket, a passenger may not have any preference of the time of the flight as long as it is on a certain day. Therefore, the tools could be refined to incorporate the effect of such criteria. The tool doesn’t make the whole decision support system. A thorough search needs to be done to build the decision support system. If the system is designed based on internet, a database of user preferences could be maintained to help the new users to answer questions. However, having “too much” or an unknown quality of information could be another problem [4].

Hence, filtering information is another potential area for future research to develop the decision support system for a distributed environment. The decision support system for a distributed environment should be adopted by online retail sellers, the car companies, or even the insurance companies to provide heir users a way of efficiently comparing between different alternatives. This type of system could also come handy to the students to choose their schools or their major. Sufficient effort and fund should be allocated to the development of this potential field of research.

Haven’t found the essay you want?