The holographic fuzzy learning for credit scoring

The holographic fuzzy classifier has been developed. Stimulus symmetrization based on optimal transformation has been implemented as a set of C functions. Stimulus-response associations are both learned and expressed in one noniterative mapping using holographic network. An initial network determines the most influential fields of input vectors. The reduced vectors are expanded to higher order statistics prior to input to the cortex cell. The classifier deals with financial, biomedical, power and geophysical problems. The loan and credit scoring application is described. Two real-world financial databases have been analyzed, one from America and the other from Europe. Classification in terms of good vs bad clients has been performed. It leads to the elimination of 88% of financial losses while reducing the good client prospective by only 13%.