Timeline for Generate data with skewed distribution and known percentiles, mean and median
Current License: CC BY-SA 3.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 26, 2015 at 9:45 | vote | accept | Matt Ellis | ||
| Mar 25, 2015 at 16:22 | comment | added | Glen_b | +1 I like this answer a lot; it gives the calculations, explains the difficulties, gives practical advice. | |
| Mar 25, 2015 at 16:18 | comment | added | whuber♦ | (+1) The risk assessment literature I have read suggests that a delta-lognormal (a mixture of a lognormal and an atom at zero) or a censored lognormal ought to be reasonable models. One way to re-interpret the question would be to attempt to fit such a distribution to the data, recognizing it will not exactly reproduce the statistics, but might fit it adequately. If a GoF test says it is reasonable, the fit could be taken as an plausible description of the population represented by the data. That's what one really wants in order to run simulations and assess uncertainties, anyway. | |
| Mar 25, 2015 at 16:12 | history | answered | Nick Cox | CC BY-SA 3.0 |