【摘要】 目的介绍一种基于机器学习的分类方法-logitBoost在判别分析中的应用。方法结合实例和模拟数据介绍了logitBoost的思想,原理,方法和步骤,就模型的拟合效果与Fisher线性判别、二次判别、logistic回归判别进行了比较,并探讨了“logitBoost判别”的优势及其在医学领域中的应用前景等问题。结果与传统方法相比,logitBoost判别在实例以及模拟数据的应用中,均显现出较好的或相似的模型预测效果。结论当传统的判别分析条件得不到满足,或判别效果不佳时,logitBoost能够达到良好的预测效果,在医学领域的判别分析中有较好的应用前景。更多还原

【Abstract】 Objective To introduce the application of LogitBoost in discriminant analysis,which is a machine-learning based technique,and discuss about the advantages and disadvantages of LogitBoost to other discriminant methods,the over-fit problem and the feature of its application in medical researches.Methods We presented the ideas and algorithms of LogitBoost,explored the application of LogitBoost with real data and simulated data sets in discriminant analysis as well as the comparison of goodness of fitting to Fisher’s linear discriminant analysis,quadratic discriminant analysis and logistic discriminant analysis.Results The results of the comparison show that ,in terms of the goodness of fitting,LogitBoost has a similar or even better effect than traditional methods.Conclusion LogitBoost displays the advantages when conditions of classical statistical discriminant techniques couldn’t be met or the predictive effect is bad,and will make a better feature of its application in medical researches.更多还原