R-squared of Random Forest predictions vs. observations

Fits a univariate random forest model y ~ x with the integer response y and the numeric, character or factor predictor x using ranger::ranger() and returns the R-squared of the observations against the predictions (see score_r2()).

Supports cross-validation via the arguments arguments cv_training_fraction (numeric between 0 and 1) and cv_iterations (integer between 1 and n) introduced via ellipsis (...). See preference_order() for further details.

Usage

f_count_rf(df, ...)

Arguments

df

(required, dataframe) with columns:

• "x": (numeric, character, factor) predictor.

• "y" (integer) counts response.

...

(optional) Accepts the arguments cv_training_fraction (numeric between 0 and 1) and cv_iterations (integer between 1 and Inf) for cross validation.

Value

numeric or numeric vector: R-squared

See also

Other preference_order_functions: f_binomial_gam(), f_binomial_glm(), f_binomial_rf(), f_categorical_rf(), f_count_gam(), f_count_glm(), f_numeric_gam(), f_numeric_glm(), f_numeric_rf(), preference_order()

Examples

data(vi_smol, package = "spatialData")

df <- data.frame(
  y = vi_smol[["vi_counts"]],
  x = vi_smol[["swi_max"]]
)

#no cross-validation
f_count_rf(df = df)
#> [1] 0.6350826

#cross-validation
f_count_rf(
  df = df,
  cv_training_fraction = 0.5,
  cv_iterations = 10
  )
#>  [1] 0.4312739 0.4366738 0.3774958 0.4673419 0.4201296 0.4269817 0.4248289
#>  [8] 0.4587598 0.4755349 0.4753168

#categorical predictor
df <- data.frame(
  y = vi_smol[["vi_counts"]],
  x = vi_smol[["koppen_zone"]]
)

f_count_rf(df = df)
#> [1] 0.7065278