How to plot a normalised cumulative histogram

I've got a vector of data of length around 5000000. I'm trying to plot a cumulative histogram of this data. I've previously plotted a normalised histogram using the trapz command:

[f,z]=hist(CharPoly,1000000);
bar(z,f/trapz(z,f))

where CharPoly is the data vector. That worked fine but I'm now looking to plot a cumulative histogram from this data and I can't figure out how to incorporate either cumsum or cumtrapz into this.

Thanks!

Answer 1

[Brendan Hamm]

If using 2014b or higher you can use the histogram command:

histogram(CharPoly,'Normalization','cdf')

If prior to 2014b use hist and bar:

[f,z]=hist(CharPoly,1000000);
% Make pdf by normalizing counts
% Divide by the total counts and the bin width to make area under curve 1.
fNorm = f/(sum(f)*(z(2)-z(1))); 
% cdf is no cumulative sum
fCDF = cumsum(fNorm);
bar(z,fCDF) % display

This is done from my head, but I believe this is correct.

[Riti Patel]

Thanks, I've tried that but the histogram turns out like this

I would have thought that, being a normalised cumulative histogram it would flatten out at 1? That's what I was aiming for as this is supposed to be modelling a probability distribution. Do you have any ideas?

[Riti Patel]

Don't worry, I've sorted it!

[Brendan Hamm]

Curiosity begs me to ask, What was the issue?