problem with polyval plotting

Hi folks,

I am trying to plot a trendline on a scattergraph but I'm not sure why the following is happening! Any help will be appreciated!

My initial code and result:

sumStart = 72;
exclude1 = cri > 30;

for i = 1 : numSamples
    allCokePercentSum(i) = sum(allCokePercent(sumStart:end, i));
    anisotropicPercentSum(i) = sum(anisotropicPercent(sumStart:end, i));
    isotropicPercentSum(i) = sum(isotropicPercent(sumStart:end, i));
    fillerPercentSum(i) = sum(fillerPercent(sumStart:end, i));
end

x_labelString = ['Histogram Percentage Sum (Lower Threshold = ' num2str(sumStart) ')'];

[allCokeFit, GOF1] = polyfit(cri, allCokePercentSum', 3);
[anisotropicFit, GOF2] = polyfit(cri, anisotropicPercentSum', 3);
[isotropicFit, GOF3] = polyfit(cri, isotropicPercentSum', 3);
[fillerFit, GOF4] = polyfit(cri, fillerPercentSum', 3);

allCokeLine = polyval(allCokeFit, cri);
anisotropicLine = polyval(allCokeFit, cri);
isotropicLine = polyval(allCokeFit, cri);
fillerLine = polyval(allCokeFit, cri);

figure;
hold on
subplot(2, 2, 1);
plot(allCokeLine, allCokePercentSum, cri, exclude1);
xlabel(x_labelString, 'FontWeight',"bold");
ylabel('CRI', 'FontWeight',"bold");
title('Greyscale Histogram Percentage Counts for "All-Coke" Threshold')

subplot(2, 2, 2);
plot(anisotropicLine, anisotropicPercentSum, cri, exclude1);
xlabel(x_labelString, 'FontWeight',"bold");
ylabel('CRI', 'FontWeight',"bold");
title('Greyscale Histogram Percentage Counts for "Anisotropic" Threshold')

subplot(2, 2, 3);
plot(isotropicLine, isotropicPercentSum, cri, exclude1);
xlabel(x_labelString, 'FontWeight',"bold");
ylabel('CRI', 'FontWeight',"bold");
title('Greyscale Histogram Percentage Counts for "Isotropic Threshold')

subplot(2, 2, 4);
plot(fillerLine, fillerPercentSum, cri, exclude1);
xlabel(x_labelString, 'FontWeight',"bold");
ylabel('CRI', 'FontWeight',"bold");
title('Greyscale Histogram Percentage Counts for "Filler" Threshold')
hold off

[dpb]

Your x values aren't sorted so they're being plotted with connecting lines in the order they are in the input arrays ...

The code is pretty convoluted and uses a lot of variables, but the idea is

[x,ix]=sort(x);     % sort the independent variable, save the sort index
y=y(ix);            % sort y same order to match
yHat=polyval(b,x);  % evaluate the fit in sorted order, too...
plot(x,y,x,yHat)    % and plot

If don't want to overwrite the original variables, can make a temporary, of course..

[Teshan Rezel]

@dpb thanks for this! may I ask what b is pelase?

[dpb]

The coefficient array from polyfit for polyval for each plot/variable...

ADDENDUM

You might find it more visually appealing to not use the line marker for the data on the plots, though...just plot the markers for data and add the regression line. plot() draws straight lines between all points in succession; that may be too "busy" or distracting...

[Teshan Rezel]

@dpb thank you! can you please post this as an answer so I can accept?

Answer 1

[dpb]

Your x values aren't sorted so they're being plotted with connecting lines in the order they are in the input arrays ...

The code is pretty convoluted and uses a lot of variables, but the idea is

[x,ix]=sort(x);     % sort the independent variable, save the sort index
y=y(ix);            % sort y same order to match
yHat=polyval(b,x);  % use coefficients, b, from polyfit to evaluate the fit in sorted order, too...
plot(x,y,x,yHat)    % and plot

If don't want to overwrite the original variables, can make a temporary, of course..

ADDENDUM

You might find it more visually appealing to not use the line marker for the data on the plots, though...just plot the markers for data and add the regression line. plot() draws straight lines between all points in succession; that may be too "busy" or distracting...

ADDENDUM SECOND

BTW, it only takes two points to draw the regression line since it is linear...

xHat=[min(x) max(x)];
yHat=polyval(b,xHat);