Tool guide

Curve Fitting: What R² Hides and How Models Compete

Curve Fitting is the regression panel of the Engineering Lab. Paste two columns of data — x and y separated by commas, spaces or tabs — choose a model family, and it fits by nonlinear least squares with the Levenberg–Marquardt algorithm. Eighteen models ship: linear, quadratic, cubic, general polynomial, exponential growth and decay, arbitrary-base exponential, double exponential, logarithmic, power law, logistic, Gompertz, exponential plateau, Michaelis–Menten, Hill, Gaussian, Lorentzian, sine and rational. Each fit returns parameters with 95 % confidence intervals, R², adjusted R², RMSE, AIC and BIC, an equation string, and a confidence band you can plot around the curve. A compare mode runs every applicable model and ranks them.

Updated Reviewed by CalcX Editorial Team5 sources5 min read