With A Little
Help From My Friends
Would you
believe in a love at first sight?
I'm certain it happens all the time, yeah
What do you see when you turn out the light?
I can't tell you, but it sure feels like mine.
How certain
are you that it happens all the time?
A random equation has two parameters: 1) its location, mean/median/mode, and
2) its scale, variance, uncertainty. The adage is that nothing is certain
except death and taxes. Given that you can cheat on taxes, but you can’t cheat
death, I would suggest that only death is certain, and thus life is uncertain, i.e. has
a variance.
An exponential distribution also has a variance but is
defined only for positive numbers. This restriction is identical to saying that
its location is zero. It still has a scale parameter, a variance, that is given
as λ. It is suggested that the exponential distribution is a distribution
of the absolute. It can be coordinate transformed by translation to any location, µ, as long as µ>0
and then its Probability Density Function, PDF, becomes
λ*e-λ*(x-µ)
and its Cumulative Distribution Function, CDF, becomes
1-e-λ*(x-µ)
The median of an exponential distribution is generally
given as ln 2/λ, but this is when the location is zero. With the translation
of the location, the median is ln 2/λ+µ. The mean is 1/λ+µ.
A normal logistics distribution has both a location and a
scale parameter. Its CDF is
½*tanh((x-µ)/(2s))+½.
For the median of the two distributions to be equal requires that s = (ln 2)/λ and that µ>0.
When the location of a logistics distribution is zero,
then its upper half, above its median, looks like an exponential distribution
with a location of zero. This is hardly surprising. The exponential distribution
is also the equation of radioactive decay. Its scale parameter, λ, is then known
as a half-life.
There is no need as Grushka
References
Grushka, E. (1972). Characteristics of Exponentially
Modifed Gaussian Peaks in Chromatograhy. Analytical Chemistry Vol 44, pp.
1733-1738.
Reyes, J., Venegas, O., & Gómez, H. W. (2018).
Exponentially-modified logistic distribution with application to mining and
nutrition data. Appl. Math 12.6, 1109-1116.