sta12_1.m

几个用于统计方法的matlab程序

load cities
whos
categories
first5=names(1:5,:)
% 用箱形图来表达
boxplot(ratings,0,'+',0)
set(gca,'YTicklabel',categories)
stdr=std(ratings);
sr=ratings./stdr(ones(329,1),:);
%寻找主成分
[pcs,newdata,variances,t2]=princomp(sr);
%列出前三个主成分
p3=pcs(:,1:3)
I=p3'*p3

plot(newdata(:,1),newdata(:,2),'+')
xlabel('1st principal component');
ylabel('2nd principal component');

gname(names)
gname
metro=[43 65 179 213 234 270 314];
names(metro,:)
rsubset=ratings;
nsubset=names;
nsubset(metro,:)=[];
rsubset(metro,:)=[];
size(rsubset)

%主成分方差
variances
%由每个主成分所解释的总方差的百分数
percent_explained=100*variances/sum(variances)
pareto(percent_explained)
xlabel('Principal Component')
ylabel('Variance Explained(%)')

%Hoteling's T2检验统计量
[st2,index]=sort(t2);         %升序排序
st2=flipud(st2);         % 降序排序的数据值
index=flipud(index);     % 降序排列的系数
extreme=index(1)
names(extreme,:)