T-HOMEWORK

代码说明：

  用Parzen窗法或者kn近邻法估计概率密度函数，得出贝叶斯分类器，对测试样本进行测试，比较与参数估计基础上得到的分类器和分类性能的差别.2． 同时采用身高和体重数据作为特征，用Fisher线性判别方法求分类器，将该分类器应用到训练和测试样本，考察训练和测试错误情况。将训练样本和求得的决策边界画到图上，同时把以往用Bayes方法求得的分类器也画到图上，比较结果的异同。3．选择上述或以前实验的任意一种方法，用留一法在训练集上估计错误率，与在测试集上得到的错误率进行比较。(Use Parzen Windows method or the kn neighbor method estimated probability density function, draw the bayes classifier, to test sample to test, and the parameter estimation is based on the classifier and get the difference of classification performance. 2. At the same time the height and weight data as the characteristic, with Fisher identification method for linear classifier, and the application of the classifier to training and test sample, investigation and test training error condition. Will the training sample and utility of decision on the drawing boundary paint, and at the same time before the Bayes method with optimal classifier also paint chart, the result of the comparison of the similarities and differences. 3. Select the above or before any a kind of method of experiment, with a method for training set in the estimation error rates, and on the test set get error rates are compared.)