Baltic Software Solutions (BSS) has introduced the Pattern recognition method (PR) for risk assessment. The idea of the method is: 1) to use the balanced scorecards for each risk class and 2) to classify the case by comparison of the calculated values of scorecards for each class. In addition, BSS has implemented the fuzzy logic and the linear programming method for the calculation of scorecard parameters (recognition training).
The pattern recognition method (PR) is a certain evolution of balance scorecard method. First of all, by introduction of several balanced scorecards instead of one, we avoid the need of definition of value ranges for the classes (decision is made by comparison of calculated values).
For the calculation of the scorecard weights values is proposed the linear programming (math). That is a step towards the revolutionizing evolution of the balanced scorecard method.
The linear programming method is widely used for general pattern recognition and is mathematically soundly based. The weight calculations are based on the given representatives for each class. The representatives are selected from the recorded past experience data. If such an experience data (history) does not exist, the method can be provided with the hypothetical case patterns, specified by the business experts.
When the pattern recognition based scoring is used, the financial business experts are concentrating on building of the most effective patterns and the best representatives of introduced classes. This fact corresponds to the financial expertsâ business nature and their skills.