FitRegularization

FitRegularization

is an option for Fit and FindFit that specifies a regularization for fitting a model.

Details

  • Fit and FindFit typically find parameters pars that minimize the norm(res) where res is the residual vector defined as the difference between the model and data response at the data coordinate points. With FitRegularization->rfun, the objective to minimize is norm(residuals)+rfun(pars)
  • Fit and FindFit find parameters that minimize the norm where is the residual vector with components given by where are the data coordinates and are the data values and model also depends on the parameters.
  • Possible settings include:
  • Noneno regularization
    rfunregularize with rfun[a]
    {"Tikhonov", λ}regularize with
    {"LASSO",λ}regularize with
    {"Variation",λ}regularize with
    {"TotalVariation",λ}regularize with
    {"Curvature",λ}regularize with
    {r1,r2,}regularize with the sum of terms from r1,
Introduced in 2019
(12.0)