how to choose the best distribution for my data

Hi guys,

I have a 2D histogram of data and would like to fit a probabilistic distribution to it. Is root mean squared error a good measure when comparing fits of different distributions ?

Answer 1

[Walter Roberson]

Have a look at kstest2(), kstest(), lillietest()

[Kurt]

hi walter,

I have already fit various distibutions using a maximum likelihood approach and computed the RMS. Please see this FEX which fits various distibution using ML and computes an RMS as well along with the fit after it has been computed. My question is : Is the RMSE provided in this FEX a good measure of the fit? http://www.mathworks.com/matlabcentral/fileexchange/4222-a-collection-of-fitting-functions

For example, for a normal distribution, see Line 84 of the following: http://www.mathworks.com/matlabcentral/fileexchange/4222-a-collection-of-fitting-functions/content/FitFunc/fit_ML_normal.m

[Kurt]

hi can anyone help me out?