Computing CDF????

Hi all,

Does anyone of you know how to calculate cdf plot using following data sets:

A = [10 10 10 5 3 3 3 2 0]; B = [10 10 10 5 5 3 3 2 2 2 1 0]; how can obtain cdf from A and B?

Best Regards Talha

Answer 1

[Steven Lord]

If you're using release R2014b or later (for histcounts and histogram) or release R2015b or later (for histcounts2 and histogram2) you can specify the 'Normalization' argument with value 'cdf' to display the CDF of the data that you binned using the appropriate function.

Note that this doesn't find parameters for a particular distribution that were determined by fitting the distribution to your data. If you want to do that, take a look at some of the functions on this page in the documentation for Statistics and Machine Learning Toolbox, specifically those whose names end in "fit".

[Camilo Malagon Nieto]

Thank you for the update. Then the solution will be:

Given....

A = [10 10 10 5 3 3 3 2 0]; B = [10 10 10 5 5 3 3 2 2 2 1 0]

First merge the data in a new vector

X=[A,B]

The plot the Cumulative distribution histogram

histogram(X,'Normalization','cdf')

Answer 2

[Image Analyst]

Try this code. It's actually just two calls - one to hist() and one to cumsum(), but it's a full-fledged demo with all kinds of fancy plotting and does it for both A and B, so it looks long and complicated but it's actually not. (It only took about 3 minutes to code up this demo). Don't be afraid - just copy, paste, and run:

clc;    % Clear the command window.
workspace;  % Make sure the workspace panel is showing.
fontSize = 20;

A = [10 10 10 5 3 3 3 2 0]; 
B = [10 10 10 5 5 3 3 2 2 2 1 0];

% Compute the histogram of A and B.
[countsA, binsA] = hist(A);
[countsB, binsB] = hist(B);

% Compute the cumulative distribution function of A and B.
cdfA = cumsum(countsA) / sum(countsA);
cdfB = cumsum(countsB) / sum(countsB);

% Plot the probability distribution of A.
subplot(2,2, 1);
bar(binsA, countsA);
title('Histogram of A', 'FontSize', fontSize);
ylabel('Count A', 'FontSize', fontSize);
xlabel('Values of A', 'FontSize', fontSize);
grid on;
% Plot the probability distribution of B.
subplot(2,2, 2);
bar(binsB, countsB);
title('Histogram of B', 'FontSize', fontSize);
ylabel('Count B', 'FontSize', fontSize);
xlabel('Values of B', 'FontSize', fontSize);
grid on;

% Plot the cumulative distribution function of A.
subplot(2,2, 3);
bar(binsA, cdfA); % You can use plot() if you want to.
title('CDF of A', 'FontSize', fontSize);
grid on;
ylabel('Percentage A (/100)', 'FontSize', fontSize);
xlabel('Values of A', 'FontSize', fontSize);
% Plot the cumulative distribution function of B.
subplot(2,2, 4);
bar(binsB, cdfB); % You can use plot() if you want to.
title('CDF of B', 'FontSize', fontSize);
ylabel('Percentage B (/100)', 'FontSize', fontSize);
xlabel('Values of B', 'FontSize', fontSize);
grid on;

% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]); % Maximize figure.
set(gcf,'name','Demo by ImageAnalyst','numbertitle','off')

Answer 3

[Talha Rahman]

Thanx Image Analyst, that is a great explanation. Let me elaborate what i was asking. I meant, is it possible to have single CDF curve from both A and B. A and B are the outcomes of same function that i ran two times and gave me A and then B. i want to know distribution of the outcomes.

Best Regards

Talha

Answer 4

[Ssssssss fffgffggg]

how to plot exp cdf with histogram?????

[Ssssssss fffgffggg]

rand(1,10000)

[Image Analyst]

This does not answer the original poster's question. To get a CDF and histogram they should call cumsum() and histogram(), not rand(). Anyway, they posted it 6 years ago so it's likely they don't need an answer anymore, despite the fact that they never accepted the answer that was offered.

Answer 5

[Camilo Malagon Nieto]

This will be my answer:

There is not a function that automatically takes a data vector and creates a vector with CDF values.

You need first to see your data by plotting it with a histogram

hist(X)

then use the distribution tool to see what distributions may suit your data

disttool

then use the distribution fitting tool to see how that distribution actually fits your data

dfittool

Then create a new distribution Object that fits your data

NewDist=fitdist(X,'Normal');

Than use the new distibution to plot the CDF

cdfplot(NewDist)

if you need values from the CDF you will have to create a values vector and the values to plot and get the CDFs for them

t=0:0,1:10;
Xcdf = cdf(NewDist,t);

Then you can plot the same

plot(t,Xcdf);

[Camilo Malagon Nieto]

Worth to notice that the data provided by the question does not have statistical meaning due to the poor amount of samples.