Matlab中的分组箱图:通用功能

时间:2014-07-15 11:49:17

标签: matlab boxplot

在SO中看到这篇精彩文章之后:

Most efficient way of drawing grouped boxplot matlab

我想知道是否有可能创建这样的函数但更通用一些,因为在我的应用程序中我需要对不同情况下的不同算法进行多次分析,调整绘图代码是非常繁琐的。每个案例。

我想要这种情节的通用:

enter image description here

1 个答案:

答案 0 :(得分:3)

我编写了一个Matlab函数,为你(我)做了这个。

特点:

  • 在每个箱图中支持不同数量的数据
  • 支持每组的任意数量的组和箱图
  • 支持Xlabel和boxplotlabel
  • 自动选择颜色或用户指定的颜色

功能结果示例:

enter image description here

CODE:

function multiple_boxplot(data,xlab,Mlab,colors)

% data is a cell matrix of MxL where in each element there is a array of N
% length. M is how many data for the same group, L, how many groups.
%
% Optional:
% xlab is a cell array of strings of length L with the names of each
% group
%
% Mlab is a cell array of strings of length M
%
% colors is a Mx4 matrix with normalized RGBA colors for each M.

% check that data is ok.
if ~iscell(data)
    error('Input data is not even a cell array!');
end

% Get sizes
M=size(data,2);
L=size(data,1);
if nargin>=4
    if size(colors,2)~=M
        error('Wrong amount of colors!');
    end
end
if nargin>=2
    if length(xlab)~=L
        error('Wrong amount of X labels given');
    end
end

% Calculate the positions of the boxes
positions=1:0.25:M*L*0.25+1+0.25*L;
positions(1:M+1:end)=[];

% Extract data and label it in the group correctly
x=[];
group=[];
for ii=1:L
    for jj=1:M
        aux=data{ii,jj};
        x=vertcat(x,aux(:));
        group=vertcat(group,ones(size(aux(:)))*jj+(ii-1)*M);
    end
end
% Plot it
boxplot(x,group, 'positions', positions);

% Set the Xlabels
aux=reshape(positions,M,[]);
labelpos = sum(aux,1)./M;

set(gca,'xtick',labelpos)
if nargin>=2
    set(gca,'xticklabel',xlab);
else
    idx=1:L;
    set(gca,'xticklabel',strsplit(num2str(idx),' '));
end


% Get some colors
if nargin>=4
    cmap=colors;
else
    cmap = hsv(M);
    cmap=vertcat(cmap,ones(1,M)*0.5);
end
color=repmat(cmap, 1, L);

% Apply colors
h = findobj(gca,'Tag','Box');
for jj=1:length(h)
   patch(get(h(jj),'XData'),get(h(jj),'YData'),color(1:3,jj)','FaceAlpha',color(4,jj));
end

if nargin>=3
    legend(fliplr(Mlab));
end
end

简单示例:

clear;clc;
% Create example data
A=rand(100,10);
B=rand(200,10);
C=rand(150,10);

% prepare data
data=cell(10,3);
for ii=1:size(data,1)
    Ac{ii}=A(:,ii);
    Bc{ii}=B(:,ii);
    Cc{ii}=C(:,ii);
end
data=vertcat(Ac,Bc,Cc);

xlab={'Hey','this','works','pretty','nicely.','And','it','has','colors','!!!!'};
col=[102,255,255, 200; 
    51,153,255, 200;
    0, 0, 255, 200];
col=col/255;

multiple_boxplot(data',xlab,{'A', 'B', 'C'},col')
title('Here it is!')

enter image description here

Mathworks文件交换文件可以在这里找到: http://www.mathworks.com/matlabcentral/fileexchange/47233-multiple-boxplot-m