Cannot find the plot options for LinearModels.fit using a constant

1 次查看（过去 30 天）

Martin Nors Pedersen 2013-2-13

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/63348-cannot-find-the-plot-options-for-linearmodels-fit-using-a-constant

I have some x,y data that may or may not show a plateau depending on experimental conditions. Where the plateau shows up on the x-axis will depend on the sample. Fitting to these plateaus will therefore be relevant for me.

I can fit to the plateaus using just a constant. My code is shown below

DataFileNumber = 11;
xdata = DataFiles{1,DataFileNumber}.data(:,1);  
ydata = DataFiles {1,DataFileNumber}.data(:,2);
% Fit model to data.
FitRange = (xdata >= 0.04) & (xdata <= 0.05);   % region with plateau
MyFit = LinearModel.fit(xdata(FitRange),ydata(FitRange), 'constant');
% Plot and display the results
disp(MyFit);
plotResiduals(MyFit,'caseorder')

The constant I get out of it and the MyFit.Fitted values corresponds well to the different plateaus when I inspect them in the variable editor. But I cannot produce a plot with the fitted constant shown together with my data. “plot(MyFit)” gives me a histogram.

I saw Richard’s video about some of the features in (Non)LinearModel.fit ( http://blogs.mathworks.com/videos/2012/03/16/regression-in-matlab/ ) and 0:48 in the video he shows a really nice plot where you can see the fitted model + std error shown together the data. Unfortunatly, you cannot see what command he uses. Does anyone know?

Generally, I cannot find the plot options for (Non) LinearModels.fit. The doc called LinearModel.plot ( http://www.mathworks.se/help/stats/linearmodel.plot.html ) list one option, plot(fitname), but Richard generates several plots. I know many of them are not relevant when I just fit a constant. But for future reference, it would be nice have a list of the commands you can use to produce plots from your fit :)