How to model a sigmoid function?

Asked Nov 13 '20 at 21:30

Active Nov 13 '20 at 21:35

Viewed 50 times

I have a variable where Y ~ sigmoid(X). Y ranges between 0 and 1. Y, however, is not a probability, so I don't know if I can just get a logistic regression. What I do want is, for a change in X, how much does Y change. How would I model this? Obviously, different values of X leads to different changes in Y, depending on where you are on the sigmoid curve.

Thanks!

edited Nov 13 '20 at 21:35

Richard Hardy

54,375
10
95
219

asked Nov 13 '20 at 21:30

confused

2,453
6
26

How would one do it? Most logistic models take 0,1 as the input for Y. It doesn't make sense for me to convert my Y to 0 and 1. – confused Nov 13 '20 at 21:37
2

Why not use beta regression? (See my answer [here](https://stats.stackexchange.com/a/29042/7290).) – gung - Reinstate Monica Nov 13 '20 at 21:45
1

I also found one here: https://stackoverflow.com/questions/33033176/using-r-to-fit-a-sigmoidal-curve – confused Nov 13 '20 at 21:56
1

Is the outcome continuous or discrete? – Glen_b Nov 14 '20 at 00:37
I suspect that the equations here might contain what you want: https://www.graphpad.com/guides/prism/latest/curve-fitting/reg_classic_dr.htm – Michael Lew Nov 14 '20 at 04:43

How to model a sigmoid function?

0 Answers0