{"id":7680,"date":"2016-07-08T09:00:43","date_gmt":"2016-07-08T13:00:43","guid":{"rendered":"https:\/\/blogs.mathworks.com\/pick\/?p=7680"},"modified":"2017-08-11T13:25:30","modified_gmt":"2017-08-11T17:25:30","slug":"alternative-box-plot","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/pick\/2016\/07\/08\/alternative-box-plot\/","title":{"rendered":"Alternative Box Plot"},"content":{"rendered":"
\r\n \r\n

Sean<\/a>'s pick this week is Alternative Box Plot<\/a> by Christopher Hummersone<\/a>.\r\n <\/p>\r\n <\/introduction>\r\n\r\n

Background<\/a><\/h3>\r\n

I was recently doing a seminar for some MATLAB users where one of the plots I showed was a boxplot<\/a><\/tt>. The box plot was used to quickly get a feel for the distributions of some energy data based on time of day. Here it is:\r\n <\/p>

figure\r\nboxplot(Power, Hour, 'Notch'<\/span>, 'on'<\/span>)\r\nxlabel('Hour of Day'<\/span>)\r\nylabel('Power (MW)'<\/span>)<\/pre> 
There is a lot of information in this plot.  The whiskers are the limits excluding outliers which are the red +s.  The boxes\r\n      represent the 25th and 75th percentiles, the red line the median, and the notches tell you if two medians are significantly\r\n      different if they don't overlap.\r\n   <\/p>\r\n   
An audience member then asked me if there was a way to control the box limits to be an arbitrary percentile, they wanted 10%\r\n      and 90%.  I searched the parameter-value pairs but did not find anything.\r\n   <\/p>\r\n   
Enter Alternative Box Plot!<\/a><\/h3>\r\n   
Alternative box plot provides many of the same options as MATLAB's boxplot but also a few more and the percentile is one of\r\n      them.\r\n   <\/p>\r\n   
Christopher has his tools in a package, which is nice to avoid naming conflicts with other functions.  I'll import<\/a><\/tt> the package to use the short name:\r\n   <\/p>
import iosr.statistics.*<\/span>\r\ny = tab2box(Hour,Power); % reshapes to be boxPlot compliant<\/span>\r\nbp = boxPlot(0:23,y,'Notch'<\/span>,true,'Percentile'<\/span>,[10 90]);<\/pre> 
Very nice!<\/p>\r\n   
You may have also noticed that I grabbed an output argument.  This allows me to customize the box plot with easy tab completion\r\n      after it's already made.\r\n   <\/p>
bp.lineColor = 'b'<\/span>;\r\nbp.medianColor = 'r'<\/span>;<\/pre> 
In fact, I could even have all four percentiles by adding additional ones.<\/p>
bp = boxPlot(0:23,y,'Notch'<\/span>,true);\r\nbp.addPrctiles = [10 90];\r\nbp.addPrctilesColors = {'r'<\/span>;'b'<\/span>}<\/pre>
bp = \r\n  boxPlot with properties:\r\n\r\n           addPrctiles: [10 90]\r\n     addPrctilesColors: {2×1 cell}\r\n     addPrctilesLabels: {2×1 cell}\r\n    addPrctilesMarkers: {2×1 cell}\r\n    addPrctilesTxtSize: []\r\n              boxAlpha: 1\r\n              boxColor: {'none'}\r\n              boxWidth: 'auto'\r\n    groupLabelFontSize: 9\r\n      groupLabelHeight: 'auto'\r\n           groupLabels: {1×0 cell}\r\n            groupWidth: 0.7500\r\n                 limit: '1.5IQR'\r\n             lineColor: {'k'}\r\n             lineStyle: {'-'}\r\n             lineWidth: 1\r\n             meanColor: {[0 0.4470 0.7410]}\r\n            meanMarker: {'+'}\r\n              meanSize: 6\r\n           medianColor: {[0 0.4470 0.7410]}\r\n                method: 'R-5'\r\n                 notch: 1\r\n            notchDepth: 0.4000\r\n        notchLineColor: {'k'}\r\n        notchLineStyle: {':'}\r\n             notchLine: 0\r\n           outlierSize: 36\r\n            percentile: [25 75]\r\n        sampleFontSize: 9\r\n            sampleSize: 0\r\n            scaleWidth: 0\r\n          scatterAlpha: 1\r\n          scatterColor: {[0.5000 0.5000 0.5000]}\r\n          scatterLayer: 'top'\r\n         scatterMarker: {'x'}\r\n           scatterSize: 36\r\n            showLegend: 0\r\n              showMean: 0\r\n          showOutliers: 1\r\n           showScatter: 0\r\n                 style: 'normal'\r\n           symbolColor: {[0 0.4470 0.7410]}\r\n          symbolMarker: {'o'}\r\n                 theme: 'default'\r\n            xSeparator: 0\r\n              xSpacing: 'x'\r\n               handles: [1×1 struct]\r\n                     x: [1×24 double]\r\n                     y: [334×24 double]\r\n               weights: [334×24 double]\r\n            statistics: [1×1 struct]\r\n<\/pre> 
Comments<\/a><\/h3>\r\n   
Give it a try and let us know what you think here<\/a> or leave a comment<\/a> for Christopher.\r\n   <\/p>