error bar on the graph

hi

after i have plot between x y and z ( scatter3(x, y, z))

x=[2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3];

y=[3.5,6.9,17.7,29.1,35.2,3.5,6.8,17.5,28.7,34.8,3.4,6.6,16.9,27.6,33.5];

z=[450,450,450,450,450,350,350,350,350,350,250,250,250,250,250];

i want to make error bar for y in the same graph

and i have the standard deviation for y

which is ( 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913)

so could you tell me what is the code for that

Answer 1

[Star Strider]

The error bars are difficult to see, so I multiplied them by 10 here:

x=[2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3];	
y=[3.5,6.9,17.7,29.1,35.2,3.5,6.8,17.5,28.7,34.8,3.4,6.6,16.9,27.6,33.5];
z=[450,450,450,450,450,350,350,350,350,350,250,250,250,250,250];
ysd = [0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913];

xr = reshape(x, [], 3);
yr = reshape(y, [], 3);
zr = reshape(z, [], 3);
ysdr = reshape(ysd, [], 3)*10;


figure
hold on
for k1 = 1:size(xr,2)
    for k2 = 1:size(xr,1)
    plot3((xr(k2,k1)*[1 1]), (yr(k2,k1)+ysdr(k2,k1)*[-1 1]), (yr(2,k1)*[1 1]), '-r')
    plot3((xr(k2,k1)), (yr(k2,k1)), (yr(2,k1)), '.r')
    end
end
hold off
view(-60,30)
grid on
xlabel('X')
ylabel('Y')
zlabel('Z')
% legend('Y\pm10\sigma')

Experiment to get the result you want.

[ahmad albngali]

thank you so much

i have anothet question related to previouse one

know i have all x values x=[2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3];

and i have y values y=[3.5,6.9,17.7,29.1,35.2,3.5,6.8,17.5,28.7,34.8,3.4,6.6,16.9,27.6,33.5];

and i have z values z=[450,450,450,450,450,350,350,350,350,350,250,250,250,250,250];

so if i have

Z = 300

x = 20

Y = ???

to bring y i used the code in the bottom

YfromXZ = scatteredInterpolant(x(:), z(:), y(:))
queryx = 20;    
queryz = 300;  
matchingy = YfromXZ(queryx, queryz);
figure;scatter3(x, y, z); hold('on'); plot3(queryx, matchingy, queryz, 'r*');

my question is

is there any code to bring the error bar for new value for Y that related to Z = 300, x = 20 ?????

[Star Strider]

As always, my pleasure!

‘my question is: is there any code to bring the error bar for new value for Y that related to Z = 300, x = 20 ?????’

There is. However that is so far from your data that I doubt the validity of the result.

I added ExtrapolationMethod to your code in order to induce it to extrapolate:

YfromXZ = scatteredInterpolant(x(:), z(:), y(:));
queryx = 20;    
queryz = 300;  
YfromXZ.ExtrapolationMethod = 'linear';
matchingy = YfromXZ(queryx, queryz);

Then the approach I took was a variation on your code:

errYfromXZ = scatteredInterpolant(x(:), z(:), y(:)+ysd(:));
errYfromXZ.ExtrapolationMethod = 'linear';
matchingyerr = errYfromXZ(queryx, queryz);
errY = (matchingyerr - matchingy)

producing:

errY =
   0.097193007483870

That is appropriately larger than the ‘ysd’ values, however even at that I am not certain I would trust it.

.

[ahmad albngali]

yes that what i want , and dont warry i will not trust it. ,

i know i am asking alo , but you rally help me

know i want to bring the error bar for both ( 1- all y values , 2- the new value for Y that related to Z = 300, x = 20 ) in one figure

i have attached an example of figure for x, y, z and alo a new value of Y when Z = 300, x = 20

so could you tell me the code that add a error bar for all y values and Y value as wll in the same figure

[Star Strider]

With the extrapolated values, reshape will not work (added points must correspond to the existing (5x3) format), so the code needs to be rearranged slightly and the extrapolated values plotted separately, after the plotting loop:

x=[2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3];	
y=[3.5,6.9,17.7,29.1,35.2,3.5,6.8,17.5,28.7,34.8,3.4,6.6,16.9,27.6,33.5];
z=[450,450,450,450,450,350,350,350,350,350,250,250,250,250,250];
ysd = [0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913];

xr = reshape(x, [], 3);
yr = reshape(y, [], 3);
zr = reshape(z, [], 3);
ysdr = reshape(ysd, [], 3)*10;

YfromXZ = scatteredInterpolant(x(:), z(:), y(:));
queryx = 20;    
queryz = 300;  
YfromXZ.ExtrapolationMethod = 'linear';
matchingy = YfromXZ(queryx, queryz);
figure;scatter3(x, y, z); hold('on'); plot3(queryx, matchingy, queryz, 'r*');

errYfromXZ = scatteredInterpolant(x(:), z(:), y(:)+ysd(:));
errYfromXZ.ExtrapolationMethod = 'linear';
matchingyerr = errYfromXZ(queryx, queryz);
errY = (matchingyerr - matchingy)

figure
hold on
for k1 = 1:size(xr,2)
    for k2 = 1:size(xr,1)
    plot3((xr(k2,k1)*[1 1]), (yr(k2,k1)+ysdr(k2,k1)*[-1 1]), (yr(2,k1)*[1 1]), '-r')
    plot3((xr(k2,k1)), (yr(k2,k1)), (yr(2,k1)), '.r')
    end
end
plot3(queryx*[1 1], matchingy+errY*[-1 1]*10, queryz*[1 1], '-g')                       % <- Added
plot3(queryx, matchingy, queryz, '.g')                                                  % <- Added
hold off
view(-60,30)
grid on
xlabel('X')
ylabel('Y')
zlabel('Z')

The extrapolated point and errorbar are plotted in green to make them more visible. (All the errorbars are multiplied by 10 because they would not otherwise be visible.)

.

[ahmad albngali]

something was wrong with z values in figher 2 in attachment

for figher 1 ( attached as well ) the z was right , but for figure 2 the all values of z from 6.6 to 6.9 and this are wrong , shold be from 250 to 450 ( as figher 1 )

and the z value for Y was right in figure 2 , because as you can see the z 300 , but all z values for x and y in figure 2 are wrong

i am tring to find why this happens , but i cant do you have any idea

[Star Strider]

When I went back and looked at my most recent code, I realize there was a typographical error I did not catch. The plot3 calls should be:

plot3((xr(k2,k1)*[1 1]), (yr(k2,k1)+ysdr(k2,k1)*[-1 1]), (zr(2,k1)*[1 1]), '-r')
plot3((xr(k2,k1)), (yr(k2,k1)), (zr(2,k1)), '.r')

I copied the ‘xr’ part of those to the z position to avoid re-typing them, and then mistyped ‘yr’ instead of ‘zr’. I was concentrating on getting the error bars to plot correctly, and overlooked that.

Fixed now. The plot3 calls after the loop are correct.

[ahmad albngali]

yes , now working great

thank you so much

is there any solution to show the error bar appear in the figure without multiplied by 10

[Star Strider]

As always, my pleasure!

The error bars are there, however they are so small with respect to the amplitudes of the other variables that they are essentially impossible to see without zooming in on each plotted data point. I multiplied them by 10 to check the code to be certain they are plotting correctly. It is not necessary to multiply them at all.

[ahmad albngali]

But if I want to put it In my thesis And I want to show people the error bar I have to multiplied by 10 Right ? That was my question is there any other cases to show the error bar without multiplied by 10 Show him the right data without multiplied by 10

[Star Strider]

Just put in the legend that the errorbars aare multiplied by 10 (or whatever number you choose), since the relative sizes may be more important than the absolute sizes, depending on what you are demonstrating. I am not certain how best to show the error bars without multiplying them.

Note that this line:

plot3((xr(k2,k1)), (yr(k2,k1)), (zr(2,k1)), '.r') 

puts a ‘.’ in the centre of the errorbars to show where it is. One possibility is to replace that by a ‘+’, although just leaving those out might work as well, so that only the error bars themselves are presented. One way to multiply them might be to use 1.96 (to show them as the 95% confidence intervals from the normal distribution) or the result of tinv since that might be more appropriate (and likely larger than 1.96), or confidence intervals based on another distribution, depending on what that distribution is (since I do not know what the data themselves represent).

[ahmad albngali]

hi as you remember i used this code

This code was used to find the Relationship of DEq, CTDIvol and scanning length, also the code was used to show the error bar for DEqvalues (The error bars for all DEqvalues are difficult to see in the graph , so I multiplied them by 5)

% x= CTDIv
x=[2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3 ];    
% y= planar average equilibrium dose (DEq)
y=[3.5,6.9,17.7,29.1,35.2,3.5,6.8,17.5,28.7,34.8,3.4,6.6,16.9,27.6,33.5 ];
% z= scanning length 
z=[450,450,450,450,450,350,350,350,350,350,250,250,250,250,250];
% ysd= DEq standard deviation 
ysd = [0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913];

xr = reshape(x, [], 3);
yr = reshape(y, [], 3);
zr = reshape(z, [], 3);
ysdr = reshape(ysd, [], 3)*5;
 
%for visualisation
figure
hold on
for k1 = 1:size(xr,2)
    for k2 = 1:size(xr,1)
    plot3((xr(k2,k1)*[1 1]), (yr(k2,k1)+ysdr(k2,k1)*[-1 1]), (zr(2,k1)*[1 1]), '-r')
    plot3((xr(k2,k1)), (yr(k2,k1)), (zr(2,k1)), '.r')
    end
end
hold off
view(-60,30)
grid on

so my question

the Error bars are too small. and i want to Increase the dot size and fit a grid or lines to the dots ??

[Star Strider]

Experiment with the 'LineWidth' and 'MarkerSize' properties to get the result you want.

See Line Properties for details.

[ahmad albngali]

Where I have to put the 'LineWidth' and 'MarkerSize' in the code Can you add it to me in the code ?

[Star Strider]

Use them in the plot3 calls. See LineWidth and MarkerSize for details.

[ahmad albngali]

hi i try to do it but it is confusing

could you please show me hoe i do it in my examble ( in my code )

[Star Strider]

Add them to the plot3 calls:

plot3((xr(k2,k1)*[1 1]), (yr(k2,k1)+ysdr(k2,k1)*[-1 1]), (zr(2,k1)*[1 1]), '-r', 'LineWidth',1.5)
plot3((xr(k2,k1)), (yr(k2,k1)), (zr(2,k1)), '.r', 'MarkerSize',10)

so the complete code is now:

% x= CTDIv
x=[2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3,2.6,5.2,14,23.3,28.3 ];    
% y= planar average equilibrium dose (DEq)
y=[3.5,6.9,17.7,29.1,35.2,3.5,6.8,17.5,28.7,34.8,3.4,6.6,16.9,27.6,33.5 ];
% z= scanning length 
z=[450,450,450,450,450,350,350,350,350,350,250,250,250,250,250];
% ysd= DEq standard deviation 
ysd = [0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913, 0.047140452, 0.047140452, 0.047140452, 0.124721913, 0.124721913];

xr = reshape(x, [], 3);
yr = reshape(y, [], 3);
zr = reshape(z, [], 3);
ysdr = reshape(ysd, [], 3)*5;
 
%for visualisation
figure
hold on
for k1 = 1:size(xr,2)
    for k2 = 1:size(xr,1)
    plot3((xr(k2,k1)*[1 1]), (yr(k2,k1)+ysdr(k2,k1)*[-1 1]), (zr(2,k1)*[1 1]), '-r', 'LineWidth',1.5)
    plot3((xr(k2,k1)), (yr(k2,k1)), (zr(2,k1)), '.r', 'MarkerSize',10)
    end
end
hold off
view(-60,30)
grid on

Change the values to create the result you want.

.