✅作者简介：热爱科研的Matlab仿真开发者，修心和技术同步精进，
代码获取、论文复现及科研仿真合作可私信。
🍎个人主页：Matlab科研工作室
🍊个人信条：格物致知。
更多Matlab完整代码及仿真定制内容点击👇
🔥 内容介绍
引言
数据可视化是数据科学中至关重要的一环。通过直观地展示数据，我们能够更好地理解数据之间的关系和趋势。在数据可视化领域，3D 散点图边缘图是一种强大的工具，它能够同时展示数据点的分布以及边缘分布的信息。本文将介绍 3D 散点图边缘图的概念、应用和优势。
概述
3D 散点图边缘图是一种用于可视化三维数据的图表类型。它通过在三维空间中绘制散点图，并在图的边缘显示数据的分布情况，提供了全面的数据展示方式。这种图表类型通常用于探索数据的分布特征、检测异常值和观察数据之间的相关性。
应用
3D 散点图边缘图在多个领域中被广泛应用。以下是一些常见的应用场景：
-
科学研究：在物理学、生物学和化学等领域，研究人员经常需要分析多维数据集。通过使用 3D 散点图边缘图，他们能够更好地理解数据点之间的关系和趋势，从而得出有关物理、生物或化学过程的重要结论。
-
金融分析：金融领域的数据通常包含多个维度，如股票价格、市场指数和交易量等。通过使用 3D 散点图边缘图，金融分析师可以更好地观察不同变量之间的关系，发现隐藏在数据中的模式和趋势。
-
地理信息系统：地理信息系统（GIS）是一种用于管理、分析和展示地理数据的工具。在 GIS 中，3D 散点图边缘图被广泛用于可视化地理数据的分布和相关性。例如，研究人员可以使用这种图表类型来分析地震数据，并确定地震活动与地理位置之间的关联。
优势
与传统的二维散点图相比，3D 散点图边缘图具有以下优势：
-
全面展示数据：3D 散点图边缘图不仅能够展示数据点的位置，还能同时展示数据的边缘分布情况。这使得我们能够更全面地理解数据的特征和趋势。
-
观察多变量之间的关系：通过在三维空间中绘制散点图，我们可以观察多个变量之间的关系。这有助于发现变量之间的相关性和模式，进而提供更深入的洞察。
-
发现异常值：3D 散点图边缘图能够帮助我们快速发现数据中的异常值。通过观察边缘分布，我们可以识别出与主要数据集有差异的数据点，从而更好地理解数据的完整性和可靠性。
结论
3D 散点图边缘图是一种强大的数据可视化工具，能够全面展示数据的分布和多变量之间的关系。它在科学研究、金融分析和地理信息系统等领域中被广泛应用，并具有展示数据、观察关系和发现异常值的优势。随着数据科学的不断发展，我们相信 3D 散点图边缘图将在更多领域中发挥重要作用，帮助我们更好地理解和利用数据的价值。
📣 代码
%% How to use scatter3mpdfclearclcclose all% data preparationN=100;dat(:,1)=linspace(0,1,N)'+0.1*randn(N,1);dat(:,2)=linspace(0,1,N)'.^2+0.1*randn(N,1);dat(:,3)=10*(0.5-linspace(0,1,N))'.^3+0.1*randn(N,1);% call function with degault option valuescatter3mpdf(dat(:,1),dat(:,2),dat(:,3))exportgraphics(gcf,'scatter3mpdf_sample.png')% call function with specified option valuefigureh=scatter3mpdf(dat(:,1),dat(:,2),dat(:,3),..."mPDFAreaRatio",0.3,..."FunctionType","cdf","Marker",'s');% set option values after plottingset(gca,'fontname','arial')colormap(cool)
function h=scatter3mpdf(x,y,z,options)%% h=scatter3mpdf(x,y,z,options)% SCATTER3MPDF is written by Eiji Konaka, Sep/2023% This function plotts the followings in one 3-d figure.% 3-d scatter plot% 2-d marginalized histogram on x-y, y-z, and z-x planes using PCOLOR% function. Relative frequency is shown by cell colors.% 1-d marginalized pdf (or cdf) along x, y, and z axes.% The plotted pdf is calculated by ksdensity, instead of histogram due to% technical reasons.%% inputs% x, y, z: data vector with the same dimensions% options% mPDFAreaRatio: The size of the ratio of pdf plot area to scatter area% (default=0.5)% FunctionType: you can select the type of distribution function from% 'pdf' or 'cdf' (default='pdf')% Marker: type of marker of scatter3 function (default='o')%% output% h: handle of figure% note% The other options, such as colormap of PCOLOR function, can not be specified% on calling this function.argumentsx (:,1) doubley (:,1) doublez (:,1) doubleoptions.mPDFAreaRatio {mustBePositive(options.mPDFAreaRatio)} =0.5options.FunctionType {mustBeMember(options.FunctionType,{'pdf','cdf'})} = 'pdf'options.Marker ='o'endscatter3(x,y,z,'blue','Marker',options.Marker);hold on;h=gcf;xLimVal=get(gca,'xlim');yLimVal=get(gca,'ylim');zLimVal=get(gca,'zlim');cMap=colormap("hot");cMap=flipud(cMap);colormap([1 1 1;cMap]);xlim([0 1]);switch options.FunctionTypecase 'pdf'[N,c] = hist3([x,y]);N_pcolor = N'/max(N,[],"all");N_pcolor(size(N_pcolor,1)+1,size(N_pcolor,2)+1) = 0;xl = linspace(min(x),max(x),size(N_pcolor,2)); % Columns of N_pcoloryl = linspace(min(y),max(y),size(N_pcolor,1)); % Rows of N_pcolor[X,Y,Z] = meshgrid(xl,yl,zLimVal(1));X=reshape(X, size(N_pcolor));Y=reshape(Y, size(N_pcolor));Z=reshape(Z, size(N_pcolor));surf(X,Y,Z,N_pcolor,'EdgeColor','none',...'FaceAlpha',0.5);[N,c] = hist3([x,z]);N_pcolor = N/max(N,[],"all");N_pcolor(size(N_pcolor,1)+1,size(N_pcolor,2)+1) = 0;xl = linspace(min(x),max(x),size(N_pcolor,2)); % Columns of N_pcolorzl = linspace(min(z),max(z),size(N_pcolor,1)); % Rows of N_pcolor[X,Y,Z] = meshgrid(xl,yLimVal(2),zl);X=reshape(X, size(N_pcolor));Y=reshape(Y, size(N_pcolor));Z=reshape(Z, size(N_pcolor));surfc(X,Y,Z,N_pcolor,'EdgeColor','none',...'FaceAlpha',0.5);[N,c] = hist3([y,z]);N_pcolor = N/max(N,[],"all");N_pcolor(size(N_pcolor,1)+1,size(N_pcolor,2)+1) = 0;yl = linspace(min(y),max(y),size(N_pcolor,2)); % Columns of N_pcolorzl = linspace(min(z),max(z),size(N_pcolor,1)); % Rows of N_pcolor[X,Y,Z] = meshgrid(xLimVal(2),yl,zl);X=reshape(X, size(N_pcolor));Y=reshape(Y, size(N_pcolor));Z=reshape(Z, size(N_pcolor));surfc(X,Y,Z,N_pcolor,'EdgeColor','none',...'FaceAlpha',0.5);[f_pdf_x,xi]=ksdensity(x);[f_pdf_y,yi]=ksdensity(y);[f_pdf_z,zi]=ksdensity(z);plot3(xi, ...(f_pdf_x-max(f_pdf_x))/(max(f_pdf_x)-min(f_pdf_x)) ...*(yLimVal(2)-yLimVal(1))*options.mPDFAreaRatio+(yLimVal(1)), ...min(zLimVal)*ones(size(xi)), ...'b-');plot3((f_pdf_y-max(f_pdf_y))/(max(f_pdf_y)-min(f_pdf_y)) ...*(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)), ...yi, ...min(zLimVal)*ones(size(xi)), ...'b-');plot3((f_pdf_z-max(f_pdf_z))/(max(f_pdf_z)-min(f_pdf_z)) ...*(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)), ...yLimVal(2)*ones(size(yi)), ...zi, ...'b-');case 'cdf'[N,c] = hist3([x,y]);tmp=N;for n1=1:size(N,1)for n2=1:size(N,2)tmp(n1,n2)=sum(sum(N(1:n1, 1:n2)));endendN=tmp;N_pcolor = N'/max(N,[],"all");N_pcolor(size(N_pcolor,1)+1,size(N_pcolor,2)+1) = 0;xl = linspace(min(x),max(x),size(N_pcolor,2)); % Columns of N_pcoloryl = linspace(min(y),max(y),size(N_pcolor,1)); % Rows of N_pcolor[X,Y,Z] = meshgrid(xl,yl,zLimVal(1));X=reshape(X, size(N_pcolor));Y=reshape(Y, size(N_pcolor));Z=reshape(Z, size(N_pcolor));surf(X,Y,Z,N_pcolor,'EdgeColor','none',...'FaceAlpha',0.5);[N,c] = hist3([x,z]);tmp=N;for n1=1:size(N,1)for n2=1:size(N,2)tmp(n1,n2)=sum(sum(N(1:n1, 1:n2)));endendN=tmp;N_pcolor = N/max(N,[],"all");N_pcolor(size(N_pcolor,1)+1,size(N_pcolor,2)+1) = 0;xl = linspace(min(x),max(x),size(N_pcolor,2)); % Columns of N_pcolorzl = linspace(min(z),max(z),size(N_pcolor,1)); % Rows of N_pcolor[X,Y,Z] = meshgrid(xl,yLimVal(2),zl);X=reshape(X, size(N_pcolor));Y=reshape(Y, size(N_pcolor));Z=reshape(Z, size(N_pcolor));surfc(X,Y,Z,N_pcolor,'EdgeColor','none',...'FaceAlpha',0.5);[N,c] = hist3([y,z]);tmp=N;for n1=1:size(N,1)for n2=1:size(N,2)tmp(n1,n2)=sum(sum(N(1:n1, 1:n2)));endendN=tmp;N_pcolor = N/max(N,[],"all");N_pcolor(size(N_pcolor,1)+1,size(N_pcolor,2)+1) = 0;yl = linspace(min(y),max(y),size(N_pcolor,2)); % Columns of N_pcolorzl = linspace(min(z),max(z),size(N_pcolor,1)); % Rows of N_pcolor[X,Y,Z] = meshgrid(xLimVal(2),yl,zl);X=reshape(X, size(N_pcolor));Y=reshape(Y, size(N_pcolor));Z=reshape(Z, size(N_pcolor));surfc(X,Y,Z,N_pcolor,'EdgeColor','none',...'FaceAlpha',0.5);[f_cdf_x, x_bins]=ecdf(x);[f_cdf_y, y_bins]=ecdf(y);[f_cdf_z, z_bins]=ecdf(z);plot3(x_bins, ...(f_cdf_x-1)*(yLimVal(2)-yLimVal(1))*options.mPDFAreaRatio+(yLimVal(1)), ...min(zLimVal)*ones(size(x_bins)), ...'b-');plot3((f_cdf_y-1)*(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)), ...y_bins, min(zLimVal)*ones(size(x_bins)), ...'b-');plot3((f_cdf_z-1)*(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)), ...max(yLimVal)*ones(size(y_bins)), z_bins,...'b-');end% 鏋犮伄杩借plot3([xLimVal(1) xLimVal(1) xLimVal(2) xLimVal(2) xLimVal(1)], ...[yLimVal(1) -(yLimVal(2)-yLimVal(1))*options.mPDFAreaRatio+(yLimVal(1)) ...-(yLimVal(2)-yLimVal(1))*options.mPDFAreaRatio+(yLimVal(1)) yLimVal(1) yLimVal(1)], ...[zLimVal(1) zLimVal(1) zLimVal(1) zLimVal(1) zLimVal(1) ],'k-' ...)plot3([-(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)) xLimVal(1) ...xLimVal(1) -(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)) -(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1))], ...[yLimVal(1) yLimVal(1) yLimVal(2) yLimVal(2) yLimVal(1)], ...[zLimVal(1) zLimVal(1) zLimVal(1) zLimVal(1) zLimVal(1) ],'k-' ...)plot3([-(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)) xLimVal(1) ...xLimVal(1) -(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1)) -(xLimVal(2)-xLimVal(1))*options.mPDFAreaRatio+(xLimVal(1))], ...[yLimVal(2) yLimVal(2) yLimVal(2) yLimVal(2) yLimVal(2)], ...[zLimVal(1) zLimVal(1) zLimVal(2) zLimVal(2) zLimVal(1) ],'k-' ...)hold offend
⛳️ 运行结果
扫一扫
1903
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
