I can't find what is the distribution corresponding to the default option in:
DistributionFitTest[data, Automatic, "HypothesisTestData"];
Is it N(0,1), uniform?
$\begingroup$
$\endgroup$
1
-
2$\begingroup$ Does this answer your question? $\endgroup$– SzabolcsMar 20, 2014 at 15:53
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
8
By default DistributionFitTest attempts to test fit against the family of normal distributions. To test if the data is standard normal (i.e. N(0,1)) you would use.
DistributionFitTest[data, NormalDistribution[]]
There are examples that show this to be the case in the Properties & Relations section of the documentation for DistributionFitTest.
-
$\begingroup$ When you have time, could you take a look whether this looks reasonable? I have an actual practical need for this. My distribution looks like a histogram, but it comes from probabilistic bisection (see pages 13-15) and not from counting data points. $\endgroup$– SzabolcsMay 12, 2014 at 20:37
-
$\begingroup$ @Szabolcs I don't see anything wrong with your approach. I don't have access to the source so I can't promise that the Infinity is safe across the board. You know the typical caveat about undocumented internals being subject to change. $\endgroup$ May 14, 2014 at 13:45
-
$\begingroup$ Suggest this may be a Beta distribution, not a normal one. Before testing the fit to any distribution, you should identify which one it is. I do this 100 times and find the best one looping FD = FindDistribution[data, 5, "BIC", "RandomSeed" -> foundi - 1, PerformanceGoal -> "Quality"] in Mathematica 10.3 The functionality may be different in other versions. $\endgroup$– CarlAug 25, 2016 at 21:26
-
$\begingroup$ @Andy Ross No, first of all ND is bounded on $(-\infty ,\infty )$ and your results as well as BD are bounded on $(0,1)$. Such that you data may be BD. Best to let Mathematica tell you which distribution it most likely is see FindDistribution. That will go through a lot of distributions and select the best ones. $\endgroup$– CarlAug 26, 2016 at 21:53
-
$\begingroup$ @Andy Ross Sorry, my bad. The results pertain to probabilistic bisection (see pages 13-15) and Szabolcs's comment. $\endgroup$– CarlAug 26, 2016 at 22:34