# Detecting the probability distribution in dataset

I have a dataset in a CSV file which is a single column of positive integers. I would like to know which probability distribution it fits.

The histogram is as follows. This is how I am trying.

w = Transpose[Import["dataset.csv"]][];
h = DistributionFitTest[w, SkewNormalDistribution[12, 7, 0.1], "HypothesisTestData"];
Show[Histogram[w, Automatic, "ProbabilityDensity"],
Plot[PDF[h["FittedDistribution"], x], {x, 0, 80},
PlotStyle -> Thick]]


The output is: Can I claim that my dataset has a skewed normal distribution?

## migrated from mathematica.stackexchange.comApr 19 '16 at 22:45

This question came from our site for users of Wolfram Mathematica.

• @MarcoB, I admit that I was confused where to post. How can I move it there? – Omar Shehab Apr 19 '16 at 16:16
• Perhaps you could ask for help from the moderators (flag your question as "in need of moderator intervention" and explain what you would like to do). Otherwise, since I see that you are already registered on Cross Validated, it may be simplest to copy and re-post there. – MarcoB Apr 19 '16 at 17:06

## 1 Answer

This is an extended comment to echo @MarcoB 's comment that this question would be better served on Cross Validated.

Your data consists of only even integers that range from 0 to 72. While you can probably find a discrete probability distribution that will produce similar observed histograms (like maybe twice a negative binomial), the repeatable patterns in the sequence of observations negates being able to talk about your data as if it was a random sample from such a distribution.

Here is a plot of the observations in sequence: Consideration of how the data is generated can be even more important that just providing a compact summary/approximation of the observed histogram.

• @Jim, what is the systematic way to find distribution in these scenarios? – Omar Shehab May 9 '16 at 21:37