I need a quantitative method of identification of which distribution best represents my data set

I have a data set. i need to find the type of distribution it follows. for example gaussian or lognormal or exponential or binomial etc. i need a quantitative method of indentification of which distribution best represents my data. I am not exactly sure how to use the kstest and adtest for different distributions. suggest if there is any other way. Thankyou.

Answer 1

[Jeff Miller]

One approach is to:

1. get the maximum likelihood parameter estimates for fitting each candidate distribution to your actual data.
2. choose the candidate distribution which gives the highest maximum likelihood

Another approach is estimate and choose based on minimizing a chi-square lack of fit measure, but this requires identifying a set of score bins into which you divide the data.