histfit using Kernel function with a specific bandwith?

Hi,

Can anyone tell me if there is way use histfit with a Kernel function, but also choose the bandwidth for the Kernel function. For example, I know I can use histfit like this:

histfit(data,10,'kernel')

...and get a nice fitted histogram with 10 bins. But, how can I change the bandwidth that is used in the Kernel function? I know how to do this using ksdensity:

[f,xi] = ksdensity(data,'width',mywidth)

...can this be applied to histfit?

Thanks in advance, Dan.

Answer 1

[Dan]

Sorry for answering my own question, but I figured it out. I just made a modified version of histfit for the kernel function distribution. I modified line 61 in histfit to:

    pd = fitdist(data,dist,'width',w);

and added the input option of w. The beauty of open source code!

[Mark van den Bosch]

Great answer!

[Francesca Lo Iacono]

Hi!

is the following line the only one you have modified?

"pd = fitdist(data,dist,'width',w);"