Dear all,
I have a question regarding to the methods used for identifying Power-law distributions. I would like to identify if my data follows Power-law distribution using graphical identification methods like Pareto quantile-quantile plots, mean residual life plots, and log–log plots. Assuming that we know how to use these methods to identify power-law distributions, do you think they can scientifically identify Power-law distributions?
I believe the paper "POWER-LAW DISTRIBUTIONS IN EMPIRICAL DATA" completely answered my question.
One thing you could do is to plot data and model using the complementary cumulative distribution and check the Xmin and exponent obtained with Clauset's method. With small datasets the maximum likelihood methods could be biased.
- Badges
- Science topic
- Similar topics
- Cognitive Science
- Thinking