Explained Deviance from univariate GAM model

Computes the explained deviance of a response against a predictor via Generalized Additive Model (GAM). This option is slower than f_rsquared(), but suitable if you will be fitting GAMs with the resulting preference order.

Usage

f_gam_deviance(x, y, df)

Arguments

x

(required, character string) name of the predictor variable.

y

(required, character string) name of the response variable

df

(required, data frame) data frame with the columns 'x' and 'y'.

Value

numeric: explained deviance

See also

Other preference_order: auc_score(), case_weights(), f_gam_auc_balanced(), f_gam_auc_unbalanced(), f_logistic_auc_balanced(), f_logistic_auc_unbalanced(), f_rf_auc_balanced(), f_rf_auc_unbalanced(), f_rf_rsquared(), f_rsquared(), preference_order()

Examples

data(vi)

#subset to limit example run time
vi <- vi[1:1000, ]

#this example requires "mgcv" installed in the system
if(requireNamespace(package = "mgcv", quietly = TRUE)){

  f_gam_deviance(
    x = "growing_season_length", #predictor
    y = "vi_mean",               #response
    df = vi
  )

}
#> [1] 0.8142463