What is the most flexible distribution type in histfit?

Hi guys,

So I need to make a plot fit from a histogram using histfit. But I can never get a curve which fits closely to my histogram. I have tried almost all distribution type but none of them give satisfactory result. For instance this is the histogram I want to add the fitting to:

In that picture I used kernel dist type. As can be seen the fitted curve should show 4 peaks but it didn't. Any idea how to solve this?

Answer 1

[José-Luis]

 doc ksdensity

Works if fitting is all you care about.

[Imam]

ksdensity is almost identical to kernel dist type in histfit. I have tried and it gave the same fit curve as above.

[José-Luis]

Then I am sorry, but I don't understand what you want to achieve. If you want the curve to look exactly like the histogram, then don't plot the frequencies as a histogram but as a line.

Alternatively, pass the points where you want the density to be evaluate to ksdensity:

a = randn(10000,1);
[f,xi] = ksdensity(a,-5:5);
plot(xi,f,'r-','LineWidth',2);
[f xi] = ksdensity(a,-5:0.1:5)
plot(xi,f,'g-','LineWidth',2);