Want help in Discretising a cummulative distribution plot.

Hi all,

Actually I want to discretize a cummulative distribution plot so that I can obtain certain bands in the cdfplot. Please see the screenshot below, it will be much helpful to visualize what i am trying to obtain. http://virgintech.org/DiscretiseCDF.JPG

Many thanks in advance!

Shahab

Answer 1

[Andrew Newell]

It sounds like what you are after is a set of quantiles. If you have a set of data for your cdf, you can use quantile.

If you have a function for your cdf such as normcdf, you would use the inverse function ( e.g., norminv).

All the above assumes you have the Statistics Toolbox.

EDIT: I looked again at your figure and realized that the independent variable is on the vertical axis (not what I would expect). To get a plot of that sort, you could do the following:

1. For each power level Pow(i), find the fraction of data less than or equal to Pow(i) to get your cumulative probability p(i).
2. Plot the bands using stair(p,Pow).

A more common approach, though, would be to select your cumulative probabilities, use quantile to determine the corresponding power levels, and plot using stair(Pow,p).

Answer 2

[Shahab Khan]

Dear Andrew, Many thanks for your quick reply. and my apologies for replying to you late. Actually I had no internet connection yesterday so couldn't reply. Btw I have the 17520 data points and from those data points I have drawn the CDF that I showed you in the link above. Are you saying that using quantile I can get those bands in CDF?

[Andrew Newell]

See my edited answer.

[Andrew Newell]

Keep in mind that responses to answers should be comments (like the one I'm writing), not answers.

[Shahab Khan]

many thanks again Andrew ...your help is appreciated!! I will try implementing your suggestion would let you know the result.

Btw i mistakenly added my response in answer section but now i will keep it in mind. Thanks

[Shahab Khan]

Hi andrew ..I tried your suggestion and it worked out :). Many thanks for your time and attention!!