2
$\begingroup$

Practioners are used to defining lognormal distributions in terms of P90 and P10 cumulative density function values. To utilize these esperts' input I need to be able to convert these P90/P10 values into parameters used in software which are mean and sigma. How do I convert P90 and P10 lognormal CDF values into the distribution's mean and sigma values

$\endgroup$
0

1 Answer 1

2
$\begingroup$

It rather depends on whether "mean and sigma" are parameters of the log-normal distribution or the underlying normal distribution. Let's assume the latter.

You take the natural logarithms of the P90 and P10 values. Their midpoint should be the mean of the underlying normal distribution, and the distance between them should be about $ 2.563$ times the standard deviation of the underlying normal distribution: i.e. $\Phi^{-1}(0.9)-\Phi^{-1}(0.1)$ based on the cumulative distribution of a standard Gaussian distribution.

$\endgroup$

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