Decision-making rules based on rough set theory: Creditworthiness case study

Accurate client (company) information is important for assessing possibility creditworthiness of a client, e.g. in the insurance industry, thus accurate information are not known in real situations. There is always uncertainty in input data which may result in inaccurate decisions. To obtain accurate decision-making rules of client creditworthiness, rough set theory was introduced to obtain knowledge rules for client creditworthiness. Attributes such as type of company, length insurance, insurance penetration, damages (percent) and liquidity (2nd degree) were combined to build a decision table. After unification (discretization and categorization) input value attributes, decision-making rules were calculated through the decision-making rule generation algorithm based on the rough set theory. A classification based on the generated rules classified the client (company) into creditworthy and uncreditworthy groups. The result of fuzzy logic was used to compare with the classification based on the rough set theory. The accuracies of the rough set (91.7%) and fuzzy logic (83.3 %) were compared. It showed the number of creditworthy client based on the rough set theory more reflects the fact (real situation). The high accuracy of creditworthy based on the rough set theory demonstrated that this method was effective for client creditworthy.