1
$\begingroup$

I would like to find a suitable distribution to fit to a dataset. Beyond visual analysis of histograms with overlaid density curves and Q-Q plots I would like to perform statistical tests, namely Kolmogorov-Smirnov and Anderson-Darling. As I don't know the fully specified distribution a priori, I estimate the distribution parameters from the data. However this invalidates the tests and so I simulate the test statistics a large number of times.

My issue is in interpreting the output. Here is an example using R code from the answer on How to determine which distribution fits my data best? to test the suitability of a Weibull distribution for my data:

library(logspline)
library(FAdist)
library(fitdistrplus)
library(ADGofTest)

n.sims <- 5e4 #number of simulations for KS and AD tests
x <- as.numeric(zooList$flow12001) #data vector length 973

fit.wei <- fitdist(x, "weibull")
#replicate KS
ksstats <- replicate(n.sims, {
  r <- rweibull(n = length(x), shape = fit.wei$estimate["shape"]
              , scale = fit.wei$estimate["scale"])
  as.numeric(ks.test(r, "pweibull", shape = fit.wei$estimate["shape"]
                     , scale = fit.wei$estimate["scale"])$statistic)      
})

ksfit <- logspline(ksstats)
kspval <- 1 - plogspline(ks.test(x, "pweibull", shape= fit.wei$estimate["shape"],
                       scale = fit.wei$estimate["scale"])$statistic, ksfit)
> kspval
[1] 0.2647569

#replicate A-D
adstats <- replicate(n.sims, {
  r <- rweibull(n = length(x), shape = fit.wei$estimate["shape"]
              , scale = fit.wei$estimate["scale"])
  as.numeric(ad.test(r, pweibull, shape= fit.wei$estimate["shape"]
                     , scale = fit.wei$estimate["scale"])$statistic)      
})

adfit <- logspline(adstats)
adpval <- 1 - plogspline(ad.test(x, pweibull, shape = fit.wei$estimate["shape"],
                       scale = fit.wei$estimate["scale"])$statistic, adfit)
> adpval
[1] 0.1292376

My interpretation of these results is that for repeated tests, the KS test results in a rejection of H0 ~26% of the time whilst the AD test a rejection of H0 ~13% of the time. Often hypothesis tests are compared to tables of critical values for the test statistic or associated p-values (generally p-val = 0.05), but in my example can I choose the p-value arbitrarily due to the resampling procedure?

I am aware that there exist tables for the critical values of the KS test (e.g. http://www.cas.usf.edu/~cconnor/colima/Kolmogorov_Smirnov.htm) and asymptotic AD test (http://www.cithep.caltech.edu/~fcp/statistics/hypothesisTest/PoissonConsistency/AndersonDarling1954.pdf) but to be honest I am not sure how to use these in a resampling-based methodology.

$\endgroup$
1
$\begingroup$

You are essentially looking at a distribution of p-values. As discussed in [1] and [2], p-value is a uniformly-distributed random variable when the null hypothesis is true. Note that in this case your null hypothesis is that your sampled data orginate from the reference distribution curve.

A good way to validate this fact is to perform a large number of K-S tests where you compare between resampled populations of your data instead of a reference distribution curve. You will see that indeed, when comparing sets of data that come from the same model, the p-value distribution is uniform.

Thus, you could perhaps use a uniformity test on the p-value distribution to determine its uniformness. If your uniformity test gives a p-value < 0.05 or some other critical value, then you can reject your null hypothesis.

[1] Murdoch, D, Tsai, Y, and Adcock, J (2008). P-Values are Random Variables. The American Statistician, 62, 242-245.

[2] Why are p-values uniformly distributed under the null hypothesis?

$\endgroup$
  • $\begingroup$ Thank you for your answer, I was not aware the p-value distribution under H0 being true is uniform. Is the KS p-value of 0.26 given above irrelevant or uninterpretable then? $\endgroup$ – boro141 Oct 19 '15 at 11:53
  • 1
    $\begingroup$ What you can say based on a single p-value of 0.26 is that the null hypothesis cannot be rejected. However, re-sampling provides an opportunity for a more robust evaluation of the hypothesis. Note that goodness-of-fit tests can be used to reject the null hypothesis with some statistical degree of confidence. However, a p-value above the critical threshold does not mean that the null hypothesis can be automatically accepted. $\endgroup$ – ikonikon Oct 19 '15 at 13:12
  • $\begingroup$ Ok thank you. Upon analysis, the distribution of the p-values does not appear to be normal according to a histogram and a further KS test (on the p-values' suitability for a uniform distribution) returning a p-value << 0.05. Thus the null hypothesis that the p-values come from a Uniform distribution can be rejected, thus the previous null hypothesis that the original data come from a Weibull distribution can be rejected. Is this last sentence correct? $\endgroup$ – boro141 Oct 19 '15 at 13:33
  • $\begingroup$ I am not sure what is meant by 'the distribution of the p-values does not appear to be normal'..do you mean 'does not appear to be uniform'? If the distribution of p-values does not look like a uniform distribution (which is the case if you get p-value << 0.05 as you say), then indeed, your last sentence is corrent. $\endgroup$ – ikonikon Oct 19 '15 at 13:46
  • $\begingroup$ Apologies, it should of course read "uniform" not "normal". Thanks $\endgroup$ – boro141 Oct 19 '15 at 13:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.