How can I divide an interval in intervals with the same length?

Hello,

I have a point cloud and I would like to do a Matlab program, which calculates the mean of values of intervals, which have the same length. I don't know how I can begin that. Please, I need your help.

Maxime

Answer 1

[Sean de Wolski]

Here's an example of what might lead you to your end goal:

A = 1:64; %vector
B = mean(reshape(A,8,8),1);  %mean of intervals 8 long.

Is this what you're looking for? If it's not please provide a small sample of input data/operation/expected output

[Maxime]

I have two 4608-by-1 vectors a and b, and the respective intervals are [2.4895e-04; 0.0177] and [0.1981; 8.9363]. I have a point cloud which is linear, when I do plot(b,a,'o'). I would like to calculate the mean of each vector in each intervals to finally plot that with standard deviation.

So, I will have a straight line with the breadth on each point and for the two vectors, which are plot.

Is it more clear?

[Sean de Wolski]

First, don't accept an answer when the problem isn't _actually_ solved. Second, no it's not clear what you want. Do you want something like errorbar ?

Perhaps you could post a picture of your plot to a file hosting website somewhere so it's a little clearer.

[Maxime]

Sorry, I clicked on the wrong touch.

Then, it is like errobar, like the second picture in the matlab help.

[Maxime]

You could see the plot here:

http://dl.free.fr/pWNQPggOe

[Sean de Wolski]

By sorting the data (to group it together) and reshaping you should be able to calculate what you want.

[Maxime]

I see. What do you think about that?

I have a 4608x1 vector a and I do

sort(a(a>2e-4 & a<4e-4)) for example and I apply this example for each interval?

[Sean de Wolski]

Actually: use histc (second output argument) and accumarray to bin-> mean the values directly.

Answer 2

[Friedrich]

Hi,

are you looking for something like this?

start_interval = 0;
end_interval  = 10;

intervals = 20;
interval_length = (end_interval - start_interval)/intervals;

x = start_interval + (end_interval-start_interval)*rand(200,1);
y = rand(200,1);

plot(x,y,'*')

hold on

for i=1:intervals
    left = start_interval + interval_length*(i-1);
    right = start_interval + interval_length*i;
    ind = x >= left & x < right;
    m_y = mean(y(ind));
    line( [left,right],[m_y,m_y], 'Color','r' );
    line( [left,left], [-0.5,1.5] ,'Color','black')
end

hold off

[Maxime]

Hi,

It is a bit like this, but instead of x and y, I have two 4608-by-1 vectors a and b, and the respective intervals are [2.4895e-04; 0.0177] and [0.1981; 8.9363]. So, how can I plot the point cloud within the mean and the standard deviation like on the errorbar tool? Or without the point cloud?

[Friedrich]

I think you have to modify my code to this

start_interval = min(a);

end_interval = max(a);

x = a;

y = b;

[Maxime]

To show the linearity, I must plot with loglog and I have a warning, which says that negative data ignored.

Then, I think that your code calculate the mean only on the values of y, and I would like to have the mean of x too.

Sorry, but do you know how can I add the standard deviation like the errorbar fonction on the plot?

[Maxime]

I am very grateful for your help and I hope there is no problem :D

Answer 3

[simar]

Probably you need to elaborate a bit more what exactly your problems is?

For this much all I can say is that you can use mean command in matlab for calculating mean of values. Use help mean to get a better insight of it.

Or

You can use this a=1 % this will represent you length of interval

b = zeros(size(X));

for n =1:length(X)-1
  b(n) =  (X(n)-X(n+1)>a)
end

z =find(b>a) % this will return all index for this is true

and then use something like this mean (X(z));

But for better understanding your problem, you must elaborate your problem.