Automatically determine the best-fitting 2D-Gaussian for a data set

autofit_gaussian_2D(
  data,
  comparison_method = "rmse",
  maxiter = 1000,
  simplify = TRUE
)

Arguments

data	A data.frame that contains the raw data (generally rectilinearly gridded data, but this is not a strict requirement). Columns must be named `"X_values"`, `"Y_values"` and `"response"`.
comparison_method	One of "rmse", "rss", or "AIC"; what metric should be used to determine the "best-fitting" Gaussian?
maxiter	Default 1000. A positive integer specifying the maximum number of iterations allowed. See `stats::nls.control()` for more details.
simplify	TRUE or FALSE. If TRUE, return only the coefficients, model, model_error_stats, and fit_method for the best-fitting model. If FALSE, a model comparison table is also included in the returned list as `$model_comparison`. This table is obtained via `compare_gaussian_fits()`.

Value

If simplify = TRUE, a list with the components:

"coefs" A data.frame of fitted model parameters.
"model" The model object, fitted by stats::nls().
"model_error_stats" A data.frame detailing the rss, rmse, deviance, and AIC of the fitted model.
"fit_method" A character vector that indicates which method and orientation strategy was used by this function.

If simplify = FALSE, a model comparison table is also included in the returned list as $model_comparison. This table is obtained via compare_gaussian_fits().

Details

This function runs fit_gaussian_2D() three times: once for each of the "main" types of models: 1) elliptical, unconstrained; 2) elliptical, log; 3) circular. In all three cases, amplitudes and orientations are unconstrained. The function compare_gaussian_fits() is then used to determine which of these three models is the best-fitting, using the comparison_method argument to make the decision.

Author

Vikram B. Baliga

Examples

if (interactive()) {
}