Developing a Genetic Fuzzy System Model for Cost-Benefit Analysis

Cost benefit analysis is a systematic approach for calculation and analyzing the cost of a project. Soft computing approaches are also applicable to deal with cost benefit analysis. In this paper Mamdani fuzzy system has been developed for cost benefit analysis. The genetic optimization of the model is carried out. The interpretability and accuracy features are also analyzed.


INTRODUCTION
Cost and completion time are the two important features of the project. Several approaches have been developed to approximate cost of the project. Cost benefit analysis is an analytical model to deal with cost approximation of the project [6]. Normally cost benefit analysis [1] have four parameters that is cost on sale (COS), quantity of sale (QOS), cost at variation (CAV), cost at fixed (CAF). To reduce the associated risk with the model different probabilistic and stochastic models have been developed.
In this paper a genetic fuzzy system has been proposed and implemented using open access software GUAJE. Fuzzy systems are applied to deal with uncertainty and imprecision existing in the applications [6,7,8]. Interpretability and Accuracy are the important features of the fuzzy systems [7,8]. They are contradicting with each other .i.e. one can be improved at the cost of other. This situation leads to interpretability-Accuracy trade off [9,10]. Fuzzy concepts are also used to data base applications [11,12]. Fuzzy logic is applied in rule base systems leading to a new area called fuzzy rule based systems (FRBS).
Genetic algorithm has been used to optimize the fuzzy system proposed for Cost benefit analysis [2]. This paper consists of 4 sections. Section 1 is related to the introduction. Section 2 is the description of proposed model. Experiments and result analysis are carried out in section 3. Section 4 is the conclusion and future scope.

Linguistic or Mamdani FRBS
In this FRBS the if -then rules have linguistic values in the consequent part of the rule, the rules are as follows: Ri : if Xi1 is Ai1 and………….and Xin is Ain then Y is Bi .

Approximate or Scatter partition FRBS
In this variable the fuzzy variables are directly used in the rules. The fuzzy if -then rules are as follows: Ri : if Xi1 is A^i1 and………….and Xin is A^in then Y is G^i1 A fuzzy system has been proposed for estimating profit in cost benefit analysis procedure. The input parameter are Cost on sale, Quantity of sale, Cost at variation, Cost at fixed. The value for these input and output parameters are tabulated below in Table 1 and Table 2 gradually.

EXPERIMENTS AND RESULT ANALYSIS
The proposed model has been implemented using tool GUAJE [4]. GUAJE stands for Generating Understandable and Accurate fuzzy models in a Java Environment. It implements the fuzzy modelling methodology named as Highly Interpretable Linguistic Knowledge (HILK) [5], which is aimed at yielding a good interpretability-accuracy trade-off thanks to combining expert and induced knowledge in a common framework. It consists on a computational environment for building interpretable and accurate fuzzy systems by means of combining several pre-existing open source tools, taking profit from the main advantages of each individual tool by analogy with the main idea underlying to Soft Computing. The data set for the proposed model is detailed in table 3. M a y 03, 2 0 1 4

Table-3 (Used data set)
During the implementation the observed results for interpretability and accuracy are as follows:

Table-4 (Interpretability measurement of corresponding data set in table-3
Here the author found the result in term of accuracy 98% and interpretability index (0.115).
The Genetic optimization on rule selection has been carried out with following parameters:

CONCLUSION & FUTURE SCOPE
A genetic fuzzy system has been implemented for the purpose of cost benefit analysis. The results of the proposed model are found satisfactory. In future the author would be interested to use interval type-2 fuzzy system for developing the proposed system.