John D. Cook • 12/20/2025

Trying to fit a logistic curve

The article discusses the difficulty of accurately fitting a logistic (S-curve) model, particularly predicting its asymptotic value, when using only data from the curve's left tail. It details a simulation experiment using Python's SciPy library, showing that curve-fitting algorithms often fail to converge with limited early data and that reliable parameter estimation requires data points beyond the inflection point.

0 comments

#simulation #Logistic Regression #Scipy