How do I plot multiple lines of best fit on a scatter plot?

Hi,

I'm struggling to plot three data series onto a scatter plot, with each series having its own line of best fit (linear). My code is below. My three data series are (x1,y1), (x1,y2) and (x2,y3) with corresponding error bars for each series. When I try plotting the lines, they do not appear correctly and also some points from the first data series do not appear on the graph, but the error bars do. I have tried reordering the code etc but I can't understand what I'm doing wrong here. Any help would be greatly appreciated.

x1 = [20 40 60 80 100] ;
x2 = [20 40 60 80] ;
y1 = [0.240671275 0.552924282 0.874381642 0.99074338 1.354697701] ;
err1 = [0.052548669, 0.08814497, 0.117517615, 0.117123163, 0.167532477] ;
y2 = [0.11101623 0.11247439 0.208469577 0.322969453 0.386224488] ;
err2 = [0.133008916 0.125080979 0.1156091 0.11321253 0.119709187] ;
y3 = [0.111780925 0.162983696 0.31700554 0.378830489] ;
err3 = [0.056024757 0.050835675 0.071003682 0.058800409] ;
figure(1)
xlim([0 105]) ;
scatter(x1,y1, 'g', 'Linewidth', 5)
Fit1 = polyfit(x1,y1,1);
plot(polyval(Fit1,x1));
eb1 = errorbar(x1,y1,err1, 'vertical', 'LineStyle', 'none') ;
set(eb1, 'color', 'k', 'LineWidth', 2);
hold on
scatter(x1,y2, 'r', 'Linewidth', 5)
Fit2 = polyfit(x1,y2,1);
plot(polyval(Fit2,x1));
eb2 = errorbar(x1,y2,err2, 'vertical', 'LineStyle', 'none') ;
set(eb2, 'color', 'k', 'LineWidth', 2)
hold on
scatter(x2,y3, 'b', 'Linewidth', 5)
Fit3 = polyfit(x2,y3,1);
plot(polyval(Fit3,x2));
eb3 = errorbar(x2,y3,err3, 'vertical', 'LineStyle', 'none') ;
set(eb3, 'color', 'k', 'LineWidth', 2);

Answer 1

[Adam Danz]

See inline comments for changes.

The most significant change is using refline() to plot the regression lines based on the fits.

x1 = [20 40 60 80 100] ;
x2 = [20 40 60 80] ;
y1 = [0.240671275 0.552924282 0.874381642 0.99074338 1.354697701] ;
err1 = [0.052548669, 0.08814497, 0.117517615, 0.117123163, 0.167532477] ;
y2 = [0.11101623 0.11247439 0.208469577 0.322969453 0.386224488] ;
err2 = [0.133008916 0.125080979 0.1156091 0.11321253 0.119709187] ;
y3 = [0.111780925 0.162983696 0.31700554 0.378830489] ;
err3 = [0.056024757 0.050835675 0.071003682 0.058800409] ;

figure(1)
hold on  % ADD THIS
xlim([0 105]) ;
scatter(x1,y1, 'g', 'Linewidth', 5)
Fit1 = polyfit(x1,y1,1);
% plot(polyval(Fit1,x1)); % REMOVE
rl(1) = refline(Fit1(1),Fit1(2));   % ADD THIS
rl(1).Color = 'g'; % ADD THIS
eb1 = errorbar(x1,y1,err1, 'vertical', 'LineStyle', 'none', 'Color', 'g', 'LineWidth', 2) ; % SET COLOR SAME AS LINE & LINE WIDTH 
% set(eb1, 'color', 'k', 'LineWidth', 2); % NO LONGER NEEDED
% hold on % NO LONGER NEEDED, MOVED TOWARD THE TOP

scatter(x1,y2, 'r', 'Linewidth', 5)
Fit2 = polyfit(x1,y2,1);
% plot(polyval(Fit2,x1));  % REMOVE
rl(2) = refline(Fit2(1),Fit2(2));   % ADD THIS
rl(2).Color = 'r'; % ADD THIS
eb2 = errorbar(x1,y2,err2, 'vertical', 'LineStyle', 'none','color', 'r', 'LineWidth', 2) ; % SET COLOR SAME AS LINE & LINE WIDTH 
% set(eb2, 'color', 'k', 'LineWidth', 2) % NO LONGER NEEDED
% hold on  % NOT NEEDED, ONLY NEEDS TO BE RUN ONCE

scatter(x2,y3, 'b', 'Linewidth', 5)
Fit3 = polyfit(x2,y3,1);
% plot(polyval(Fit3,x2)); % REMOVE
rl(3) = refline(Fit3(1),Fit3(2));   % ADD THIS
rl(3).Color = 'b'; % ADD THIS
eb3 = errorbar(x2,y3,err3, 'vertical', 'LineStyle', 'none', 'color', 'b', 'LineWidth', 2) ;% SET COLOR SAME AS LINE & LINE WIDTH 
% set(eb3, 'color', 'k', 'LineWidth', 2); % NOT NEEDED, ONLY NEEDS TO BE RUN ONCE

[Joshua Chaplin]

Very grateful for this Adam, thanks!

Answer 2

[dpb]

All of

Fit1 = polyfit(x1,y1,1);
plot(polyval(Fit1,x1));

should be

Fit1 = polyfit(x1,y1,1);
plot(x1,polyval(Fit1,x1));

You plotted the predicted value versus ordinal number of the observation, not the x vector of interest.

Better would be to compute and save the predicted values and plot them, probably. Then you've got those values for error, etc.

Fit1 = polyfit(x1,y1,1);
yHat1=polyval(Fit1,x1));
hL1=plot(x1,yHat1);