《数学建模算法与应用》(第三版)课件 第7章 数理统计.pptx

第7章;目录 CONTENTS;参数估计和 假设检验;7.1.1 区间估计;clc, clear x0=[506 508 499 503 504 510 497 512 514 505 493 496 506 502 509 496]; x0=x0(:); pd = fitdist(x0,Normal) %对数据进行正态分布拟合 ci = paramci(pd,Alpha,0.05) %ci的第一列是均值的置信区间 [mu,s,muci,sci] = normfit(x0,0.05) %另一种方法求置信区间;clc, clear alpha=0.05; ta=tinv(1-alpha,4) x0=[1050 1100 1120 1250 1280]; xb=mean(x0), s=std(x0), n=length(x0); mu=xb-s/sqrt(n)*ta %计算单侧置信下限 [h,p,ci]=ttest(x0,xb,Alpha,0.05,Tail,right) %通过假设检验求置信区间;clc, clear x1=[6.683, 6.681, 6.676, 6.678, 6.679, 6.672]; x2=[6.661, 6.661, 6.667, 6.667, 6.664]; [h1,p1,ci1,st1]=ttest(x1,mean(x1),Alpha,0.1) %均值检验和区间估计 [h2,p2,ci2,st2]=ttest(x2,mean(x2),Alpha,0.1) [h3,p3,ci3,st3]=vartest(x1,var(x1),Alpha,0.1) %方差检验和区间估计 [h4,p4,ci4,st4]=vartest(x2,var(x1),Alpha,0.1);clc, clear x1=[6.683, 6.681, 6.676, 6.678, 6.679, 6.672]; x2=[6.661, 6.661, 6.667, 6.667, 6.664]; [h,p,ci,st]=ttest2(x1,x2,Alpha,0.1) ?;7.1.2 经验分布函数;clc, clear, close all a=readmatrix(data7_5_1.txt); a=a(~isnan(a)); %去掉NaN值 [f,x]=ecdf(a) %计算经验分布函数的取值 ecdf(a) %画经验分布函数图形 grid on, xlabel($x$,Interpreter,Latex) ylabel($F(x)$,Interpreter,Latex) writematrix([x,f], data7_5_2.xlsx);7.1.3 Q-Q图;clc, clear, close all a=readmatrix(data7_5_1.txt); a=a(~isnan(a)); %去掉NaN值 pd=fitdist(a, Normal), qqplot(a,pd) %Matlab工具箱直接画Q-Q图 sa=sort(a); %把a按照从小到大排列 n=length(a); pi=([1:n]-1/2)/n; yi=norminv(pi,pd.mu,pd.sigma); %计算对应的yi值 hold on, plot(yi,sa,o) %再重新描点画Q-Q图;7.1.4 非参数检验;clc, clear, n=100; bins=[0:7]; mi=[36 40 19 2 0 2 1 0]; pd = fitdist(bins,Poisson,Frequency,mi) expCounts = n * pdf(pd,bins) %计算期望的频数 [h,p,st] = chi2gof(bins,Ctrs,bins,Frequency,mi, ... Expected,expCounts,NParams,1) k2 = chi2inv(0.95, st.df) %求临界值,st.df为自由度;clc, clear edges=[0:100:300 inf]; bins=[50 150 250 350]; %定义原始数据区域的边界和中心 mi=[121 78 43 58]; %已知观测频数 pd=makedist(exp,200) %定义指数分布 expect=sum(mi)*diff(cdf(pd, edges)) %计算期望频数 [h,p,st]=chi2gof(bins,Edges,edges,cdf,pd,Frequency,mi) k2=chi2inv(0.95,st.df)