Minimum-Bayes-classifier-error-rate

代码说明：

  这是模式识别中最小错误率Bayes分类器设计方案。 自行完善了在不同先验概率的条件下，男、女错误率和总错误率的统计，放入各个数组当中。 全部程序由主函数、最大似然估计求取概率密度子函数、最小错误率贝叶斯分类器决策子函数三块组成。 调用最大似然估计求取概率密度子函数时，第一步获取样本数据，存储为矩阵；第二步对矩阵的每一行求和，并除以样本总数N，得到平均值向量；第三步是应用公式（3-43）采用矩阵运算和循环控制语句，求得协方差矩阵；第四步通过协方差矩阵求得方差和相关系数，从而得到概率密度函数。 调用最小错误率贝叶斯分类器决策子函数时，根据先验概率数组，通过比较概率大小判断一个体重身高二维向量代表的人是男是女。 主函数第一步打开“MAIL.TXT”和“FEMALE.TXT”文件，并调用最大似然估计求取概率密度子函数，对分类器进行训练。第二步打开“test2.txt”，调用最小错误率贝叶斯分类器决策子函数，然后再将数组中逐一与已知性别的数据比较，就可以得到不同先验概率条件下错误率的统计。 (This is the minimum error rate pattern recognition Bayes classifier design. Self- improvement prior probability in different conditions , male , female and total error rate error rate statistics , into which each array . All programs from the main function , maximum likelihood estimation subroutine strike probability density , the minimum error rate Bayesian classifier composed of decision-making three subfunctions . Strike called maximum likelihood estimate probability density subroutine , the first step to obtain the sample data , stored as a matrix the second step of the matrix, each row sum , and divided by the total number of samples N, be the average vector third step is to application of the formula ( 3-43 ) using matrix and loop control statements , obtain the covariance matrix fourth step through the variance-covariance matrix and correlation coefficient obtained , resulting in the probability density function . Call the minimum error rate decision Functions Bayesian)