A Bayesian discriminant analysis method based on the proportional odds model is pro-posed,which combines the advantages of the proportional odds model and Bayesian statistics,and can be used for the modelling and prediction of ordered multi-classification problems.Statistical simula-tions show that the improved discriminant method and the traditional discriminant method have com-parable discriminant abilities when the overall data conform to normal distribution,otherwise,the pro-posed new method can better discriminate different data features.Meanwhile,the new method is ap-plied to analyse the real data set,which proves the effectiveness and superiority of the method in clas-sification prediction.
关键词
密度比模型/贝叶斯判别分析法/比例优势模型/经验似然
Key words
density ratio model/bayesian discriminant analysis/proportional odds model/empirical likelihood
维普
万方数据