How do I do a linear fit on half of the data points in a set?

I have a code that should plot a set of data points and then a linear fit to the data points. Only the second half of the data follows a linear trend so I have set the fit over the second half of the data only. I have the code

fig3 = figure;
for ij = 1:N
    subplot(1,2,1);
    plot(Pdissarray(ij,:)',peakposarray(ij,2:end)','.','color', CM(ij,:))
    xlabel('Dissipated Power (mW)');
    ylabel('Peak Wavelength (nm)');
    ylim([842 855]);
    hold on
    fitA = polyfit(Pdissarray(ij,15:end),peakposarray(ij,15:end),1);
    lambdafit = polyval(fitA,Pdiss(ij,:));
    subplot(1,2,1);
    plot(Pdissarray(ij,:),lambdafit,'--','color',CM(ij,:));
    hold on
end

where pdissarray and peakposarray are data sets such that each row is the thing I want to plot for each iteration of the loop and there are N rows (i.e. the arrays are each Nx50 and each N represents a single data set). The plot of these data sets is great, I get a set of N lines where each is a plot of the first row of each array. The problem is the fit. I'm trying to do a polynomial fit of degree one on each one of these lines. I do polyfit followed by polyval and then plot this output but the plot of this fit is empty. What am I doing wrong? Is it an indexing problem?

[madhan ravi]

first upload your data(values)

[JJH]

I'm attaching the files. Each row represents one data set such that the first row of Pdiss should be plotted with the first row of peakpos. I realised there is a typo in the code above:

    fitA = polyfit(Pdissarray(ij,15:end),peakposarray(ij,15:end),1);

should be

    fitA = polyfit(Pdissarray(ij,15:end),peakposarray(ij,16:end),1);

but this is not the problem with the data fitting.

Answer 1

[Star Strider]

There are several problems with the code you posted.

This works for me:

fig3 = figure;
for ij = 1:N
    subplot(1,N,ij);
    plot(Pdissarray(ij,:)',peakposarray(ij,2:end)','.','color', CM(ij,:))
    xlabel('Dissipated Power (mW)');
    ylabel('Peak Wavelength (nm)');
    ylim([842 855]);
    hold on
    fitA = polyfit(Pdissarray(ij,15:end),peakposarray(ij,16:end),1);
    lambdafit = polyval(fitA,Pdissarray(ij,:));
    plot(Pdissarray(ij,:),lambdafit,'--','color',CM(ij,:));
    hold off
end

[JJH]

Hi, thanks for your help, this is close to what I want, but I'd like all the plots to be on the same figure. Is that in how you defined subplot?

[Star Strider]

My pleasure.

You included a subplot call, so I assumed you wanted them all as subplots. To plot them all in one axes object, this works:

fig3 = figure;
hold all
for ij = 1:N
    plot(Pdissarray(ij,:)',peakposarray(ij,2:end)','.','color', CM(ij,:))
    fitA = polyfit(Pdissarray(ij,15:end),peakposarray(ij,16:end),1);
    lambdafit = polyval(fitA,Pdissarray(ij,:));
    plot(Pdissarray(ij,:),lambdafit,'--','color',CM(ij,:));
end
hold off
xlabel('Dissipated Power (mW)');
ylabel('Peak Wavelength (nm)');
ylim([845 853]);

Keeping only the operations affected by the for looop inside it makes it more efficient. I also tweaked the ylim argument to make the plot a bit more readable.