【摘要】 分类在许多领域都是重要问题,弹性判别分析是有效解决多类问题的分类方法.基于纸币的四个属性,应用这个方法来鉴别纸币的真伪.运用统计软件R可以得到:训练集的最高准确率达到99.64%,对应检测集的错误率为0.73%.同时,由于回归模型的多样性,弹性判别分析有多种形式.根据数据的特征可以提出适当的方法,对于纸币鉴别的数据集,最好的弹性判别分析的形式应用了带有适应选择项和样条光滑参数的加性模型.更多还原

【Abstract】 Classification is a important problem in many fields.Flexible discriminant analysis is classified method which solves multigroup problems efficiently.This article applies this method to classify the banknote based on 4 attributes.By means of R statistical software,the calculation states that the highest accuracy of training dataset reaches up to 99.64% and the error rate of test dataset is 0.73%.Furthermore,flexible discriminant analysis has many forms on account of the variety of regression models.It’s feasible to come up with adaptive method depending on the feature of data.For the dataset of banknote authentication,the best form of flexible discriminant analysis applies the additive models with adaptive selection of terms and spline smoothing parameters.更多还原