Corrections to Be Addressed on The Fuzzy Relation Model.
1. Implement mathematical formulae before hierarchical fusion i.e. provide theory behind the mathematical formulae of the Repayment Capability, Credit Worthiness, Credit Reputation and Credibility fuzzy relations
2. Your membership functions need to come out clearly i.e. provide the equations giving those plots of the mfs. (note based on experiments it has been concluded that the three membership functions triangular, trapezoid, and Gaussian provide the same result. For decision making case three of membership functions are equally good. But for taking parameter values as required from the fuzzy relations/ mathematical formulae values, Gaussian membership function may be easier than triangular.)
3. Rule base generation issues i.e. rules should be generated from the fuzzy relations mathematical formulae. Use standard methods to generate the rules. E.g. Products can be got using Larsen, Mamdani, TSK, etc.
4. Fuzzy model generalization i.e. should not be tied to the data collected.
5. Testing of the model i.e. provide mechanism to test the accuracy of the model consider using confusion matrix or any other well know testing approach.
6. Mention the underlying theories behind the model i.e. clearly indicate the underlying theories behind the fuzzy relation model on the write-up
The files that I will provide for the work
1. Fis files for Repayment capability, credit worthiness, credit reputation and credibility
2. Write up
3. Data set
The minimum requirements expected:
1. Edit the fis files for Repayment capability, credit worthiness, credit reputation and credibility to address corrections 1, 2, 3, 4 and 5.
2. Create a Simulink for the new Fuzzy Relation Model.
3. Edit the sections of shared write up to include the corrections 1 to 6 above. Enable track changes for easy review.