Standard deviation, Mean values for each class

Hello, I have a problem with a code.

First of all I have a .csv file with 2 columns, as I show you in the Image. in the first column the data are split by 0.5 (1,1.5,2).

I would like to find mean values and standard deviation for each of one class, I mean one value of standard deviation and mean value for the data with 1 in , one value of standard deviation for the 1.5 and one value of standard deviation for the 2.

I mean I wanna use the errorbar command.

I think I should use a loop who reads 1st column one by one but I can not make it.

Could anyone help me?

Answer 1

[Adam Danz]

Use grpstats() if you have the Stats & Machine Learning Toolbox.

data = [ 1 1 1.5 1.5 2 2; 10 9 7 6 3.5 8]',;
[dataMean, dataStd] = grpstats(data(:,2), data(:,1), {'mean', 'std'})

If your data are in table format,

Tstats = grpstats(T, 'data1', {'mean','std'}); 
%                     ^^^^^ VariableName for column 1

If you don't have Stats & Machine Learning Toolbox,

stats = splitapply(@(x)[mean(x), std(x)], data(:,2), findgroups(data(:,1))); 

[Ivan Mich]

Thank you firt of all.

I am trying tu run your suggested commands but in command window returns me the following message:

Attempt to execute SCRIPT std as a function:

I don't know why is this happening.

could you help me?

[Adam Danz]

std is a builtin Matlab function. However, you have a script (m-file) named "std.m" so when you try to use the std() function, Matlab is accessing your std.m script instead.

Solution: rename the std.m file. Avoid naming matlab files after commonly used functions. You may need to rehash your path afterwards using rehash path.

To find your std.m file, execute: which std -all

[Ivan Mich]

ok,

One more question. Do you know how to plot thw errorbars in a plot. I mean i would like to make a plot with xvaluew, y values from my .csv data, and in the same plot i would like to show the errorbars..

[Adam Danz]

Check out these errorbar examples. If you have trouble applying them to your data, let me know what the trouble is. https://www.mathworks.com/help/matlab/ref/errorbar.html#bvc1i63-8

[Ivan Mich]

the problem is that I would like to plot my final plot like this

but I can't make it. This plot is from an another software.

in x axis there are the values in column 1, y values are in column 2 .

from the command

stats = splitapply(@(x)[mean(x), std(x)], data(:,2), findgroups(data(:,1)));

you suggested are coming out 2 columns each of them has 3 values.

Do you know how to make it?

[Adam Danz]

Regarding the errorbar question, it's really easy.

1. Use the errorbar function to plot the main line and errorbars
2. Add the values from the main line to the error values to produce the dotted lines.

% lower error line
plot(x, y-err)
% upper error line
plot(x, y+err)

Regarding the splitapply output, the rows are the groups (1, 1.5, 2) and the columns are [mean, std].

[Ivan Mich]

well, I send you the full code in order to see it and tell me where I am wrong:

clc
clear

filename1= 'e.xlsx'
[d1,tex]= xlsread(filename1);
x=d1(:,1);
y=d1(:,2);

stats = splitapply(@(y)[mean(y), std(y)], y, findgroups(x)); 

meanvalues=stats(:,1)
stdvalues=stats(:,2)

 err=errorbar(meanvalues,stdvalues)
 % lower error line
 hold on
 plot(x,y,'bo')
 hold on
plot(x, y-err)
hold on
% upper error line
plot(x, y+err)

Command window send me the message :

Undefined operator '-' for input arguments of type 'matlab.graphics.chart.primitive.ErrorBar'.

Error in eo (line 24)

[Adam Danz]

plot(x, y-err)
plot(x, y+err)

1) You should be using the meanvalues insetad of x.

2) err, in your code, is the handle to the error bars. You should be using the stdvals.