Abstract:

The authors proposed a new mixed classifier which is based on affinity propagation(AP) clustering algorithm and support vector machine(SVM). Using affinity propagation clustering algorithm which optimizes data set, we will get the highquality, small sample training data for SVM classifier, which resolves the problem of support vector machine classification inaccuracy. Experimental results show that the mixed model obtains higher classification accuracy and is better than traditional support vector machine. Especially applied to predicting heart disease, the proposed classifier obtains better results.

Key words: support vector machine, affinity propagation clustering algorithm, mixed classifier, prediction of heart disease