Changing Histogram to PDF

clear all
close all
clc

M=10000;
N=100000;

figure(1)
for ii=1:N
  x =sqrt(12)*(rand(1,M) -0.5);  
  if ii==1
      mx = mean (x)
      sx2 = var(x)
      subplot(2,1,1), plot(x)
      subplot(2,1,2), hist(x,50)
  end
  y(ii) = sum(x)/sqrt(M);
end

mean (y)
var(y)
figure(2)
subplot(2,1,1), plot(y)
subplot(2,1,2), hist(y,50)

-------------------------------------------------

I want to make Gaussian PDF from this histogram.

Please help me!!!!!!

Answer 1

[Adam Danz]

Use histogram instead of hist.

Then use histcounts along with the pdf option to get the pdf.

h = histogram(y,50);
p = histcounts(y,50,'Normalization','pdf');

% plot it
figure
binCenters = h.BinEdges + (h.BinWidth/2);
plot(binCenters(1:end-1), p, 'r-')

[SL: fixed typo]

[Steven Lord]

Why not do it with histogram alone?

y = randn(1, 1e5);
h = histogram(y, 50, 'Normalization', 'pdf');

If you need it to be a smooth(er, depending on how many bins you have) curve, rather than bars:

ycoords = h.Values;
edgecoords = h.BinEdges;
xcoords = (edgecoords(1:end-1)+edgecoords(2:end))./2;
hold on
plot(xcoords, ycoords)

[Adam Danz]

As far as I can tell, the only difference is the pdf line can be plotted without first plotting the historgram bars if the histcounts method is used. If the histogram bars are desired, then using histogram() directly would be more efficient.