Dynamic Supervised Learning: Some Basic Issues and Application Aspects

This paper discusses ideas for adaptive learning which can capture dynamic aspects of real-world datasets. Although some of these ideas have a general character and could be applied to any supervised algorithm, here we focus attention on the linear, logistic and quadratic discriminant. The classifiers use Statistical Process Control (SPC) to appropriately update the rule or modify it by modifying the “training data”. These methods are tried out on simulated data and real data from the credit industry.