How to write matlab program for histogram with out using hist function.

    Xw = abs(fft(Xt));
    ki = hist(Xw,L)%%%L=bin lenth,Xw=1000 randam samples

Answer 1

[Image Analyst]

Hint: use a for loop over all the elements in your array. Convert the value of the array into a bin number. Then increment the "counts" array at that bin number:

for k = 1 : length(Xw(:))
    binNumber = ...... (you do this part)
    counts(binNumber) = counts(binNumber) + 1;
end
bar(counts, 'FaceColor', 'b');

Answer 2

[Steven Lord]

Easy. Call HISTCOUNTS or HISTOGRAM instead of HIST.

Answer 3

[rodrigo figueiredo]

a[]=0;
for p =1 : ?? :length(Xt)
for k = 1 : floor(abs(fft(Xt)))
    a= [a Xt]
end
end 
nbins = 25;
h = histogram(a,nbins)

[Walter Roberson]

That code is a highly inefficient way of replicating the original Xt signal many times, and then taking a histogram of the repeated signal -- which is going to have the same relative portions as the original signal would have.

However, the original poster wants to take a histogram of the fft output, rather than having the mean() of the signal determine how many copies of the original signal to make.

Answer 4

[Walter Roberson]

N = 1000;
L = 50;  %number of bins, not bin width!
Xt = randn(1,N) + atan(pi*rand(1,N));

Xw = abs(fft(Xt));
figure
hist(Xw,L);

minXw = min(Xw);
maxXw = max(Xw);
binwidth = (maxXw - minXw)/L;

binnum = 1 + floor((Xw - minXw) / binwidth);
bincounts = accumarray(binnum(:), 1);
X = minXw + (0:L) * binwidth;
figure
bar(X, bincounts)

bar has minor differences in the default bar width; it would not be difficult to adjust.