MATLAB | 如何使用MATLAB绘制高度自定义的桑基图(sankey)

news/2024/11/20 21:19:20/

我之前也出过一个超简单的桑基图绘制函数,但是无法应对很多特殊情况,在这里我将其重构了一些写成了类,加了很多内置修饰函数,实现了流入流出数据不相等或者跨层数据流动的特殊情况绘制,首先展示一下使用我编写的函数能够实现怎样的效果吧:

以上的图片全是本文的示例,照旧先讲解咋用,工具函数放在文末吧!!!


用法介绍

0 数据输入

数据输入要求三列元胞数组,分别代表Source流向的起点,Target流向的终点及,Value就是流动的数据量:

links={'a1','A',1.2;'a2','A',1;'a1','B',.6;'a3','A',1; 'a3','C',.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','AA',1.5; 'C','BB',2.3; 'C','AA',1.2};

1 基本绘制

其实主要就两行即可绘制:

links={'a1','A',1.2;'a2','A',1;'a1','B',.6;'a3','A',1; 'a3','C',0.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','AA',1.5; 'C','BB',2.3; 'C','AA',1.2};% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 开始绘图(Start drawing)
SK.draw()


2 连接渲染方式

连接渲染方式有:

  • ‘left’
  • ‘right’
  • ‘interp’(default)
  • ‘map’
  • ‘simple’

几种其中渐变渲染interp是默认渲染方式,修改渲染方式可以通过在draw之前设置RenderingMethod属性进行,例如(还是之前的数据):

% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 修改链接颜色渲染方式(Set link color rendering method)
% 'left'/'right'/'interp'(default)/'map'/'simple'
SK.RenderingMethod='left';  % 开始绘图(Start drawing)
SK.draw()

left

即连接颜色与左侧节点方块相同:

right

interp

map

即依据Value大小进行上色,可通过colormap函数改颜色条。

simple

纯灰色:


3 方块对齐方式

就是分别’up’/‘down’/‘center’(default)设置上对齐、下对齐、居中,通过设置Align进行配置,依旧需要draw之前设置:

% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 修改对齐方式(Set alignment)
% 'up'/'down'/'center'(default)
SK.Align='up';% 开始绘图(Start drawing)
SK.draw()

up

down

center


4 文本位置

文本位置无非就是上下左右居中:‘left’(default)/‘right’/‘top’/‘center’/‘bottom’,通过LabelLocation属性进行设置:

% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 修改文本位置(Set Text Location)
% 'left'(default)/'right'/'top'/'center'/'bottom'
SK.LabelLocation='top';SK.Sep=.2;% 开始绘图(Start drawing)
SK.draw()

注意位置设置到top及down时,为了避免遮挡可以设置Sep调整方块间空隙占比:

left

right

top

center

bottom


5 间隙与占比

方块横向宽度占比可以通过设置BlockScale属性设置要大于0小于1,越大方块占比越大。

% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 设置方块占比(Set the scale of blocks)
% BlockScale>0 & BlockScale<1
SK.BlockScale=.4;% 开始绘图(Start drawing)
SK.draw()

如果调成.001:

同时调一下上下空隙占比:

% 设置缝隙占比(Separation distance proportion)
SK.Sep=.4;


6 配色设置

通过设置ColorList属性更改预设配色:

% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 设置颜色(Set color)
SK.ColorList=[0.46, 0.54, 0.46;0.54, 0.68, 0.46;0.41, 0.49, 0.36;0.38, 0.53, 0.84;0.44, 0.59, 0.87;0.58, 0.79, 0.93;0.65, 0.64, 0.84;0.63, 0.63, 0.80;0.56, 0.53, 0.67;0.76, 0.81, 0.43;0.56, 0.86, 0.97;0.78, 0.59, 0.65;0.89, 0.91, 0.53;0.93, 0.56, 0.25;];% 开始绘图(Start drawing)
SK.draw()

当然可以只改部分配色(比如把a3改为黑色):

links={'a1','A',1.2;'a2','A',1;'a1','B',.6;'a3','A',1; 'a3','C',0.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','AA',1.5; 'C','BB',2.3; 'C','AA',1.2};% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 设置颜色(Set color)
SK.ColorList(3,:)=[0,0,0];% 开始绘图(Start drawing)
SK.draw()


7 流入及流出数据不相等

流入及流出数据不相等这样特殊的图也能画啦,原本的数据把C到BB数据改小成0.5 (links{16,3}=.5):

% 流入及流出数据不相等(Unequal inflow and outflow data)
links={'a1','A',1.2;'a2','A',1;'a1','B',.6;'a3','A',1; 'a3','C',0.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','AA',1.5; 'C','BB',2.3; 'C','AA',1.2};
links{16,3}=.5;% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 开始绘图(Start drawing)
SK.draw();


8 跨层级流动

% 含跨层级流动(Including cross level flow)
links={'a1','A',1.2;'a2','A',2;'a1','B',.6;'a3','D',1; 'a3','C',0.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','D',1.5; 'C','BB',2.3; 'C','AA',1.2;'D','AA',1.4; 'D','BB',1.1};% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));SK.Sep=.1;% 开始绘图(Start drawing)
SK.draw()

这样直接画会有遮挡,还会有比较别扭的连接次序(下面的连接往上面连,上面的连接往下面连)

默认的连接次序是

unique([Source;Target],'stable')

这样计算的,当然这个属性可以自己设置,我们先调整连接的总顺序:

% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 修改节点排列次序(Modify node arrangement order)
SK.NodeList={'a3','a1','a2','b1','b2','b3','c1','c2','c3','D','A','B','C','AA','BB'};SK.Sep=.1;% 开始绘图(Start drawing)
SK.draw()

顺眼多了,但是放在最中间还是挡着,我们想将其往上移动,可以使用moveBlockY函数,这里D节点是第10个节点,我们将其向上移动6个单位值(若是Value都很大就多移动点),注意因为要移动的是已经画好的东西,所以移动的操作要放在draw操作后:

% 含跨层级流动(Including cross level flow)
links={'a1','A',1.2;'a2','A',2;'a1','B',.6;'a3','D',1; 'a3','C',0.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','D',1.5; 'C','BB',2.3; 'C','AA',1.2;'D','AA',1.4; 'D','BB',1.1};% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 修改节点排列次序(Modify node arrangement order)
SK.NodeList={'a3','a1','a2','b1','b2','b3','c1','c2','c3','D','A','B','C','AA','BB'};SK.Sep=.1;% 开始绘图(Start drawing)
SK.draw()% 修改节点Y轴位置变化(Modify the position change of node Y direction)
SK.moveBlockY(10,+6);


9 已绘制完成图形修饰

修饰方块就用setBlock(n,prop,…)的形式,例如(将2号方块加黑色粗轮廓线):

因为要修饰的是已经画好的东西,所以修饰的操作要放在draw操作后:

links={'a1','A',1.2;'a2','A',1;'a1','B',.6;'a3','A',1; 'a3','C',0.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','AA',1.5; 'C','BB',2.3; 'C','AA',1.2};% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 开始绘图(Start drawing)
SK.draw()% 设置方块属性(Set Block Properties)
SK.setBlock(2,'EdgeColor',[0,0,0],'LineWidth',6)

循环都修饰一下(全设置为灰色):

% 循环设置方块属性(Loop Set Block Properties)
for i=1:14SK.setBlock(i,'FaceColor',[.5,.5,.5])
end

设置连接弦就使用setLink(n,prop,…)

% 设置连接属性(Set Link Properties)
SK.setLink(5,'FaceColor',[0,0,0],'FaceAlpha',.5)

设置文本setLabel(n,prop,…)

% 设置标签属性(Set Label Properties)
SK.setLabel(11,'FontSize',40,'Color',[0,0,.8])

顺嘴一提,添加标题还是title函数就可以:

title(gca,'sankey plot by slandarer','FontSize',30,'FontName','Cambria')


10 环形桑基图

本文函数可以和我写的有向弦图函数联动搞出环形桑基图,有向弦图的函数biChordChart可在fileexchange或者gitee仓库获取,篇幅问题不再赘述:

biChordChart 有向弦图 fileexchange

Zhaoxu Liu / slandarer (2023). Digraph chord chart 有向弦图 (https://www.mathworks.com/matlabcentral/fileexchange/121043-digraph-chord-chart), MATLAB Central File Exchange. 检索来源 2023/4/1.

gitee仓库

https://gitee.com/slandarer/matlab-chord-chart

未经许可代码不能做任何商务用途,引用可以引用上述fileexchange链接,引用时可以可改变引用文本格式,但至少应包含链接。

联动使用代码

links={'a1','A',1.2;'a2','A',1;'a1','B',.6;'a3','A',1; 'a3','C',0.5;'b1','B',.4; 'b2','B',1;'b3','B',1; 'c1','C',1;'c2','C',1;  'c3','C',1;'A','AA',2; 'A','BB',1.2;'B','BB',1.5; 'B','AA',1.5; 'C','BB',2.3; 'C','AA',1.2};% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));
SK.draw();close allfigure('Name','sankey demo6','Units','normalized','Position',[.05,.05,.59,.8])
BCC=biChordChart(SK.AdjMat,'Arrow','on','Label',SK.NodeList);
BCC.CData=[[65,140,240;252,180,65;224,64,10;5,100,146;191,191,191;26,59,105;255,227,130;18,156,221;202,107,75;0,92,219;243,210,136;80,99,129;241,185,168;224,131,10;120,147,190]./255;[127,91,93;187,128,110;197,173,143;59,71,111;104,95,126;76,103,86;112,112,124;72,39,24;197,119,106;160,126,88;238,208,146]./255];
BCC=BCC.draw();% 添加刻度
BCC.tickState('on')
BCC.tickLabelState('on')BCC.setTickFont('FontName','Cambria','FontSize',11)
BCC.setFont('FontName','Cambria','FontSize',17)BCC.setLabelRadius(1.32);


11 附加小案例(一)

需要注意的是这里节点是中文,务必设置字体为支持中文的字体,例如宋体。

clc;clear;
links{7,3}='';
for i=1:7links{i,1}=['浏览',num2str(i)];links{i,2}=['浏览',num2str(i+1)];links{i,3}=10000-1400*i;
end
for i=1:7links{i+7,1}=['浏览',num2str(i)];links{i+7,2}=['下载',num2str(i)];links{i+7,3}=900;
end
for i=1:7links{i+14,1}=['浏览',num2str(i)];links{i+14,2}=['流失',num2str(i)];links{i+14,3}=500;if i>=3links{i+14,3}=1100;end
end
for i=1:6links{i+21,1}=['下载',num2str(i)];links{i+21,2}=['浏览',num2str(i+2)];links{i+21,3}=600;
end
for i=1:6links{i+27,1}=['下载',num2str(i)];links{i+27,2}=['流失',num2str(i+1)];links{i+27,3}=300;
end% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));SK.NodeList={'浏览1','浏览2','浏览3','浏览4','浏览5','浏览6','浏览7','浏览8',...'下载1','下载2','下载3','下载4','下载5','下载6','下载7',...'流失1','流失2','流失3','流失4','流失5','流失6','流失7'};
SK.ColorList=[197,141,91;69,168,134;114,191,220;193,135,146;242,132,98;249,190,89;207,202,100;171,203,110;repmat([114,158,158],[7,1]);repmat([100,136,177],[7,1])]./255;% 修改对齐方式(Set alignment)
% 'up'/'down'/'center'(default)
SK.Align='down';% 修改链接颜色渲染方式(Set link color rendering method)
% 'left'/'right'/'interp'(default)/'map'/'simple'
SK.RenderingMethod='left'; % 修改文本位置(Set Text Location)
% 'left'(default)/'right'/'top'/'center'/'bottom'
SK.LabelLocation='right';% 设置方块占比(Set the scale of blocks)
% BlockScale>0 & BlockScale<1
SK.BlockScale=.16;% 开始绘图(Start drawing)
SK.draw()% 循环设置标签属性(Loop Set Label Properties)
for i=1:22SK.setLabel(i,'FontName','宋体','FontSize',12)
endfor i=10:15SK.moveBlockY(i,(9-i).*1000);
end
for i=17:22SK.moveBlockY(i,(16-i).*1000);
end

三种对齐方式的效果:

实际上不进行任何额外参数调整的基础版也挺好看的:

% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));SK.NodeList={'浏览1','浏览2','浏览3','浏览4','浏览5','浏览6','浏览7','浏览8',...'下载1','下载2','下载3','下载4','下载5','下载6','下载7',...'流失1','流失2','流失3','流失4','流失5','流失6','流失7'};% 修改对齐方式(Set alignment)
% 'up'/'down'/'center'(default)
SK.Align='top';% 修改链接颜色渲染方式(Set link color rendering method)
% 'left'/'right'/'interp'(default)/'map'/'simple'
SK.RenderingMethod='left'; % 修改文本位置(Set Text Location)
% 'left'(default)/'right'/'top'/'center'/'bottom'
SK.LabelLocation='right';% 开始绘图(Start drawing)
SK.draw()% 循环设置标签属性(Loop Set Label Properties)
for i=1:22SK.setLabel(i,'FontName','宋体','FontSize',12)
end


12 附加小案例(二)

% 随机生成数据(Randomly generated data)
clc;clear;
SourceValue=randi([1,30],[1,9]);
LayerNum=[9,6,4,7,10];
links{1,3}='';
for k=1:4TargetValue=zeros(1,LayerNum(k+1));for i=1:LayerNum(k)tValue=randi([0,13],[1,LayerNum(k+1)]);tValue=tValue./sum(tValue).*SourceValue(i);for j=1:LayerNum(k+1)TargetValue(j)=TargetValue(j)+tValue(j);if tValue(j)>epstLen=size(links,1);links{tLen+1,1}=[char(64+k),num2str(i)];links{tLen+1,2}=[char(64+k+1),num2str(j)];links{tLen+1,3}=tValue(j);endendendSourceValue=TargetValue;
end
links(1,:)=[];% 创建桑基图对象(Create a Sankey diagram object)
SK=SSankey(links(:,1),links(:,2),links(:,3));% 修改链接颜色渲染方式(Set link color rendering method)
% 'left'/'right'/'interp'(default)/'map'/'simple'
SK.RenderingMethod='interp';  % 修改对齐方式(Set alignment)
% 'up'/'down'/'center'(default)
SK.Align='center';% 修改文本位置(Set Text Location)
% 'left'(default)/'right'/'top'/'center'/'bottom'
SK.LabelLocation='top';% 设置缝隙占比(Separation distance proportion)
SK.Sep=.4;% 开始绘图(Start drawing)
SK.draw()

可自行调整渲染方式:


工具函数完整代码

classdef SSankey < handle
% Copyright (c) 2023, Zhaoxu Liu / slandarer
% =========================================================================
% @author : slandarer
% 公众号  : slandarer随笔
% 知乎    : slandarer
% -------------------------------------------------------------------------
% Zhaoxu Liu / slandarer (2023). sankey plot 
% (https://www.mathworks.com/matlabcentral/fileexchange/128679-sankey-plot), 
% MATLAB Central File Exchange. 检索来源 2023/4/28.propertiesSource;Target;Value;SourceInd;TargetInd;Layer;LayerPos;AdjMat;BoolMat;RenderingMethod='interp'  % 'left'/'right'/'interp'/'map'/'simple'LabelLocation='left'      % 'left'/'right'/'top'/'center'/'bottom'Align='center'            % 'up'/'down'/'center'BlockScale=0.05;          %  BlockScale>0 ! !Sep=0.05;                 %  Sep>=0 ! !NodeList={};ColorList=[[65,140,240;252,180,65;224,64,10;5,100,146;191,191,191;26,59,105;255,227,130;18,156,221;202,107,75;0,92,219;243,210,136;80,99,129;241,185,168;224,131,10;120,147,190]./255;[127,91,93;187,128,110;197,173,143;59,71,111;104,95,126;76,103,86;112,112,124;72,39,24;197,119,106;160,126,88;238,208,146]./255];BlockHdl;LinkHdl;LabelHdl;ax;Parent;BN;LN;VN;TotalLen;SepLen;arginList={'RenderingMethod','LabelLocation','BlockScale',...'Sep','Align','ColorList','Parent','NameList'}end
% 构造函数 =================================================================methodsfunction obj=SSankey(varargin)% 获取基本数据 -------------------------------------------------if isa(varargin{1},'matlab.graphics.axis.Axes')obj.ax=varargin{1};varargin(1)=[];else  endobj.Source=varargin{1};obj.Target=varargin{2};obj.Value=varargin{3};varargin(1:3)=[];% 获取其他信息 -------------------------------------------------for i=1:2:(length(varargin)-1)tid=ismember(obj.arginList,varargin{i});if any(tid)obj.(obj.arginList{tid})=varargin{i+1};endendif isempty(obj.ax)&&(~isempty(obj.Parent)),obj.ax=obj.Parent;endif isempty(obj.ax),obj.ax=gca;endobj.ax.NextPlot='add';% 基本数据预处理 -----------------------------------------------if isempty(obj.NodeList)obj.NodeList=[obj.Source;obj.Target];obj.NodeList=unique(obj.NodeList,'stable');endobj.BN=length(obj.NodeList);if length(obj.NodeList)>size(obj.ColorList,1)obj.ColorList=[obj.ColorList;rand(length(obj.NodeList),3).*.7];endobj.VN=length(obj.Value);% 坐标区域基础设置 ---------------------------------------------obj.ax.YDir='reverse';obj.ax.XColor='none';obj.ax.YColor='none';end
% Copyright (c) 2023, Zhaoxu Liu / slandarer
% =========================================================================
% @author : slandarer
% 公众号  : slandarer随笔
% 知乎    : slandarer
% -------------------------------------------------------------------------
% Zhaoxu Liu / slandarer (2023). sankey plot 
% (https://www.mathworks.com/matlabcentral/fileexchange/128679-sankey-plot), 
% MATLAB Central File Exchange. 检索来源 2023/4/28.
% 绘图函数 =================================================================function draw(obj)% 生成整体邻接矩阵 ---------------------------------------------obj.AdjMat=zeros(obj.BN,obj.BN);for i=1:length(obj.Source)obj.SourceInd(i)=find(strcmp(obj.Source{i},obj.NodeList));obj.TargetInd(i)=find(strcmp(obj.Target{i},obj.NodeList));obj.AdjMat(obj.SourceInd(i),obj.TargetInd(i))=obj.Value{i};endobj.BoolMat=abs(obj.AdjMat)>0;% 计算每个对象位于的层、每层方块长度、每个方块位置 ----------------obj.Layer=zeros(obj.BN,1);obj.Layer(sum(obj.BoolMat,1)==0)=1;startMat=diag(obj.Layer);for i=1:(obj.BN-1)tLayer=(sum(startMat*obj.BoolMat^i,1)>0).*(i+1);obj.Layer=max([obj.Layer,tLayer'],[],2);endobj.LN=max(obj.Layer);obj.TotalLen=max([sum(obj.AdjMat,1).',sum(obj.AdjMat,2)],[],2);obj.SepLen=max(obj.TotalLen).*obj.Sep;obj.LayerPos=zeros(obj.BN,4);for i=1:obj.LNtBlockInd=find(obj.Layer==i);tBlockLen=[0;cumsum(obj.TotalLen(tBlockInd))];tY1=tBlockLen(1:end-1)+(0:length(tBlockInd)-1).'.*obj.SepLen;tY2=tBlockLen(2:end)+(0:length(tBlockInd)-1).'.*obj.SepLen;obj.LayerPos(tBlockInd,3)=tY1;obj.LayerPos(tBlockInd,4)=tY2;% for j=1:length(tY2)%     plot([i,i],[tY1(j),tY2(j)],'LineWidth',2)% endendobj.LayerPos(:,1)=obj.Layer;obj.LayerPos(:,2)=obj.Layer+obj.BlockScale;% 根据对齐方式调整Y坐标 -----------------------------------------tMinY=min(obj.LayerPos(:,3));tMaxY=max(obj.LayerPos(:,4));for i=1:obj.LNtBlockInd=find(obj.Layer==i);tBlockPos3=obj.LayerPos(tBlockInd,3);tBlockPos4=obj.LayerPos(tBlockInd,4);switch obj.Aligncase 'up'case 'down'obj.LayerPos(tBlockInd,3)=obj.LayerPos(tBlockInd,3)+tMaxY-max(tBlockPos4);obj.LayerPos(tBlockInd,4)=obj.LayerPos(tBlockInd,4)+tMaxY-max(tBlockPos4);case 'center'obj.LayerPos(tBlockInd,3)=obj.LayerPos(tBlockInd,3)+...min(tBlockPos3)/2-max(tBlockPos4)/2+tMinY/2-tMaxY/2;obj.LayerPos(tBlockInd,4)=obj.LayerPos(tBlockInd,4)+...min(tBlockPos3)/2-max(tBlockPos4)/2+tMinY/2-tMaxY/2;endend% 绘制连接 -----------------------------------------------------for i=1:obj.VNtSource=obj.SourceInd(i);tTarget=obj.TargetInd(i);tS1=sum(obj.AdjMat(tSource,1:(tTarget-1)))+obj.LayerPos(tSource,3);tS2=sum(obj.AdjMat(tSource,1:tTarget))+obj.LayerPos(tSource,3);tT1=sum(obj.AdjMat(1:(tSource-1),tTarget))+obj.LayerPos(tTarget,3);tT2=sum(obj.AdjMat(1:tSource,tTarget))+obj.LayerPos(tTarget,3);if isempty(tS1),tS1=0;endif isempty(tT1),tT1=0;endtX=[obj.LayerPos(tSource,1),obj.LayerPos(tSource,2),obj.LayerPos(tTarget,1),obj.LayerPos(tTarget,2)];qX=linspace(obj.LayerPos(tSource,1),obj.LayerPos(tTarget,2),200);qT=linspace(0,1,50);qY1=interp1(tX,[tS1,tS1,tT1,tT1],qX,'pchip');qY2=interp1(tX,[tS2,tS2,tT2,tT2],qX,'pchip');XX=repmat(qX,[50,1]);YY=qY1.*(qT'.*0+1)+(qY2-qY1).*(qT');MeshC=ones(50,200,3);switch obj.RenderingMethodcase 'left'MeshC(:,:,1)=MeshC(:,:,1).*obj.ColorList(tSource,1);MeshC(:,:,2)=MeshC(:,:,2).*obj.ColorList(tSource,2);MeshC(:,:,3)=MeshC(:,:,3).*obj.ColorList(tSource,3);case 'right'MeshC(:,:,1)=MeshC(:,:,1).*obj.ColorList(tTarget,1);MeshC(:,:,2)=MeshC(:,:,2).*obj.ColorList(tTarget,2);MeshC(:,:,3)=MeshC(:,:,3).*obj.ColorList(tTarget,3);case 'interp'MeshC(:,:,1)=repmat(linspace(obj.ColorList(tSource,1),obj.ColorList(tTarget,1),200),[50,1]);MeshC(:,:,2)=repmat(linspace(obj.ColorList(tSource,2),obj.ColorList(tTarget,2),200),[50,1]);MeshC(:,:,3)=repmat(linspace(obj.ColorList(tSource,3),obj.ColorList(tTarget,3),200),[50,1]);case 'map'MeshC=MeshC(:,:,1).*obj.Value{i};case 'simple'MeshC(:,:,1)=MeshC(:,:,1).*.6;MeshC(:,:,2)=MeshC(:,:,2).*.6;MeshC(:,:,3)=MeshC(:,:,3).*.6;endobj.LinkHdl(i)=surf(obj.ax,XX,YY,XX.*0,'EdgeColor','none','FaceAlpha',.3,'CData',MeshC);end% 绘制方块 -----------------------------------------------------for i=1:obj.BNobj.BlockHdl(i)=fill(obj.ax,obj.LayerPos(i,[1,2,2,1]),...obj.LayerPos(i,[3,3,4,4]),obj.ColorList(i,:),'EdgeColor','none');end% 绘制文本 -----------------------------------------------------for i=1:obj.BNswitch obj.LabelLocationcase 'right'obj.LabelHdl(i)=text(obj.ax,obj.LayerPos(i,2),mean(obj.LayerPos(i,[3,4])),...[' ',obj.NodeList{i}],'FontSize',15,'FontName','Times New Roman','HorizontalAlignment','left');case 'left'obj.LabelHdl(i)=text(obj.ax,obj.LayerPos(i,1),mean(obj.LayerPos(i,[3,4])),...[obj.NodeList{i},' '],'FontSize',15,'FontName','Times New Roman','HorizontalAlignment','right');case 'top'obj.LabelHdl(i)=text(obj.ax,mean(obj.LayerPos(i,[1,2])),obj.LayerPos(i,3),...obj.NodeList{i},'FontSize',15,'FontName','Times New Roman','HorizontalAlignment','center','VerticalAlignment','bottom');case 'center'obj.LabelHdl(i)=text(obj.ax,mean(obj.LayerPos(i,[1,2])),mean(obj.LayerPos(i,[3,4])),...obj.NodeList{i},'FontSize',15,'FontName','Times New Roman','HorizontalAlignment','center');case 'bottom'obj.LabelHdl(i)=text(obj.ax,mean(obj.LayerPos(i,[1,2])),obj.LayerPos(i,4),...obj.NodeList{i},'FontSize',15,'FontName','Times New Roman','HorizontalAlignment','center','VerticalAlignment','top');endend% -------------------------------------------------------------axis tight;help SSankeyend
% Copyright (c) 2023, Zhaoxu Liu / slandarer
% =========================================================================
% @author : slandarer
% 公众号  : slandarer随笔
% 知乎    : slandarer
% -------------------------------------------------------------------------
% Zhaoxu Liu / slandarer (2023). sankey plot 
% (https://www.mathworks.com/matlabcentral/fileexchange/128679-sankey-plot), 
% MATLAB Central File Exchange. 检索来源 2023/4/28.
% =========================================================================function setBlock(obj,n,varargin)set(obj.BlockHdl(n),varargin{:})endfunction setLink(obj,n,varargin)set(obj.LinkHdl(n),varargin{:})endfunction setLabel(obj,n,varargin)set(obj.LabelHdl(n),varargin{:})endfunction moveBlockY(obj,n,dy)obj.LayerPos(n,[3,4])=obj.LayerPos(n,[3,4])-dy;set(obj.BlockHdl(n),'YData',obj.LayerPos(n,[3,3,4,4]));switch obj.LabelLocationcase 'right',set(obj.LabelHdl(n),'Position',[obj.LayerPos(n,2),mean(obj.LayerPos(n,[3,4]))]);case 'left',set(obj.LabelHdl(n),'Position',[obj.LayerPos(n,1),mean(obj.LayerPos(n,[3,4]))]);case 'top',set(obj.LabelHdl(n),'Position',[mean(obj.LayerPos(n,[1,2])),obj.LayerPos(n,3)]);case 'center',set(obj.LabelHdl(n),'Position',[mean(obj.LayerPos(n,[1,2])),mean(obj.LayerPos(n,[3,4]))]);case 'bottom',set(obj.LabelHdl(n),'Position',[mean(obj.LayerPos(n,[1,2])),obj.LayerPos(n,4)]);endfor i=1:obj.VNtSource=obj.SourceInd(i);tTarget=obj.TargetInd(i);if tSource==n||tTarget==ntS1=sum(obj.AdjMat(tSource,1:(tTarget-1)))+obj.LayerPos(tSource,3);tS2=sum(obj.AdjMat(tSource,1:tTarget))+obj.LayerPos(tSource,3);tT1=sum(obj.AdjMat(1:(tSource-1),tTarget))+obj.LayerPos(tTarget,3);tT2=sum(obj.AdjMat(1:tSource,tTarget))+obj.LayerPos(tTarget,3);if isempty(tS1),tS1=0;endif isempty(tT1),tT1=0;endtX=[obj.LayerPos(tSource,1),obj.LayerPos(tSource,2),obj.LayerPos(tTarget,1),obj.LayerPos(tTarget,2)];qX=linspace(obj.LayerPos(tSource,1),obj.LayerPos(tTarget,2),200);qT=linspace(0,1,50);qY1=interp1(tX,[tS1,tS1,tT1,tT1],qX,'pchip');qY2=interp1(tX,[tS2,tS2,tT2,tT2],qX,'pchip');YY=qY1.*(qT'.*0+1)+(qY2-qY1).*(qT');set(obj.LinkHdl(i),'YData',YY);endendendend
% Copyright (c) 2023, Zhaoxu Liu / slandarer
% =========================================================================
% @author : slandarer
% 公众号  : slandarer随笔
% 知乎    : slandarer
% -------------------------------------------------------------------------
% Zhaoxu Liu / slandarer (2023). sankey plot 
% (https://www.mathworks.com/matlabcentral/fileexchange/128679-sankey-plot), 
% MATLAB Central File Exchange. 检索来源 2023/4/28.
end

本代码编写及案例选取真的不易,希望大家该点赞的点赞,该在看的在看!!

未经允许本代码请勿作商业用途,引用的话可以引用我file exchange上的链接,可使用如下格式:

Zhaoxu Liu / slandarer (2023). sankey plot (https://www.mathworks.com/matlabcentral/fileexchange/128679-sankey-plot), MATLAB Central File Exchange. 检索来源 2023/4/28.

若转载请保留以上file exchange链接及本文链接!!!!!

该工具可通过上述fileexchange链接获取,或者通过以下gitee仓库下载:

https://gitee.com/slandarer/matlab-sankey-diagram


http://www.ppmy.cn/news/81973.html

相关文章

Linux下的用户分类与su/sudo 命令,Linux下的文件类型/用户文件权限身份/文件权限属性/权限与文件权限/ls-l文件属性详解

Tips 下载就是把我们的文件拷贝到系统的某个特定路径之下&#xff0c;普通用户是不允许你往系统里面去拷的。 Linux下的用户分类 root用户&#xff0c;管理员级别的用户身份&#xff0c;他的话基本上不受权限的约束。普通用户&#xff0c;普通用户的添加与每个普通用户密码的…

微信小程序富文本组件mp-html

功能介绍 支持在多个主流的小程序平台和 uni-app 中使用支持丰富的标签&#xff08;包括 table、video、svg 等&#xff09;支持丰富的事件效果&#xff08;自动预览图片、链接处理等&#xff09;支持设置占位图&#xff08;加载中、出错时、预览时&#xff09;支持锚点跳转、…

BetaFlight Mark4之“妖怪”声音之二

BetaFlight Mark4之“妖怪”声音之二 1. 源由2. 回顾3. 分析4. 优化5. 数据6. 综述7. 参考资料8. 附录9. 补充&#xff1a;存在yaw方向上等频率间隔gyro干扰是什么引起的&#xff1f;电机振动&#xff0c;机架共振&#xff1f;10. 补充&#xff1a;电机除了160Hz附近一个高峰以…

Redis的五大数据类型

一、String数据类型 概述&#xff1a;String是redis最基本的类型&#xff0c;最大能存储512MB的数据&#xff0c;String类型是二进制安全的&#xff0c;即可以存储任何数据、比如数字、图片、序列化对象等 1. SET/GET/APPEND/STRLEN: append命令&#xff1a;append key valu…

亏损?盈利?禾赛科技Q1财报背后的激光雷达赛道「现实」

随着禾赛科技在去年登陆美股&#xff0c;作为全球为数不多已经开始前装量产交付的激光雷达上市公司&#xff0c;财务数据的变化&#xff0c;也在一定程度上反映了行业的真实状况。 根据禾赛科技最新发布的今年一季度财报显示&#xff0c;公司季度净营收为4.3亿元&#xff08;人…

【数据结构】从数据结构角度深入探究队列

队列是计算机科学中的一种基本数据结构&#xff0c;用于存储和管理数据。在计算机程序中&#xff0c;队列被广泛应用于任务调度、进程管理等场景。本文将介绍队列的概念、特点、常见操作以及应用。 文章目录 队列的概念队列的应用队列的存储结构队列接口的实现队列的初始化队尾…

从传统开发到低代码:这是一次技术革命

近年来&#xff0c;随着人工智能等新兴技术的快速发展&#xff0c;软件开发行业也在不断演变。传统的软件开发流程需要大量的编程知识和时间&#xff0c;而且往往需要复杂的架构和开发环境。然而&#xff0c;随着低/无代码平台的出现&#xff0c;软件开发变得更加高效、简单和普…

电表及配电监控系统的智能化发展

安科瑞虞佳豪 在电力领域&#xff0c;AI渗透率近年来也持续提升。今年3月&#xff0c;国家能源局发布《关于加快推进能源数字化智能化发展的若干意见》&#xff0c;文件中针对电力、、油气等行业数字化智能化转型&#xff0c;明确了指导思想和基本原则&#xff0c;从加快行业转…