Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.

I have a logistic growth curve as follows:

$y = \frac{1}{(1 + ae^{-bx})}$, where x is the independent measure (x-axis) and a and b are paramaters. The inflection point of this equation is when y = 0.5.

Given that information, I need to find the x-value at y = 0.5. But this equation isn't easy to solve. I'm using the statistical language R to find these fits.

So, given the parameters a and b, how can I find the x-value at y = 0.5 (and again, I can't do things like derivatives because I am using a programming language). By the way, I did try to solve for x, but it's quite ugly and I'm not sure how to do it...

share|improve this question
Hint: Writing the formula in the location-scale form $y = (1 + \exp(-b(x - \log(a)/b)))^{-1}$ shows that if $x_0$ is the solution when $\log(a)=0$ and $b=1$, then $x_0 + \log(a)/b$ is the general solution. Can you find $x_0$ such that $1/2 = y = 1/(1 + \exp(-x_0))$? – whuber Oct 12 '12 at 14:58

1 Answer

up vote 5 down vote accepted

Solving for x, you get:

x = (-1/b) * ln( (1-y) / (a*y) )

Insert y = 0.5 and you get:

x = (-1/b) * ln(1/a)

It isn't that ugly... Practice the math - I promise it will help you :)

share|improve this answer
ln(-1/a)??? but "a" is positive. – CodeGuy Oct 12 '12 at 1:59
@CodeGuy then that point is not on the plot. Look closely: If y equals 0.5, then (1-a*e^(-bx))=2, meaning a*e^(-bx)=-1 right? now e^(-bx) is always positive, so a has to be negative always. – Bitwise Oct 12 '12 at 2:15
oh sorry...the original function should be 1 + a... – CodeGuy Oct 12 '12 at 2:36
@CodeGuy I changed the question and answer accordingly. – Bitwise Oct 12 '12 at 10:36

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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