ggpartialdep: Plot of partial dependence

In NaiaraLopezRojo/globalIRmap: What the Package Does (One Line, Title Case)

Description

Produce line or raster ggplots of univariate or bivariate partial dependence. i.e., estimates of the marginal relationship between predictor variables and the model’s predictions (probability of intermittence) by holding the rest of the predictors at their respective mean values. Bivariate plots show the co-linearity in response between two predictors.

Usage

ggpartialdep(
  in_rftuned,
  in_predvars,
  colnums,
  ngrid,
  nodupli = T,
  nvariate = 2,
  parallel = T,
  spatial_rsp = FALSE
)

Arguments

in_rftuned	Output from selecttrain_rf; list containing inner and outer resampling results + task.
in_predvars	data.table of predictor variable codes, names and attributes. Output from selectformat_predvars.
colnums	number of variables to include, in decreasing order of variable importance.
ngrid	number of predictor variable values to check model's marginally predicted value for.
nodupli	whether to include variable types only once (i.e. if minimum discharge is in, not including mean discharge)
nvariate	(1 or 2) whether to analyze univariate (1) or bivariate (2) partial dependence.
parallel	(boolean) whether to compute function in parrallel (as it can be computationally intensive).
spatial_rsp	(boolean) whether to use outputs from spatial (TRUE) or non-spatial (FALSE) cross-validation.

Details

this function is used to produce Figure S5 in the Supplementary Information for Messager et al. 2021. This function was initially developed for bivariate plots but only univariate plots were produced for the final manuscript so there may be obsolete snippets left.

Value

pages of gridded plots with each page containing 9 plots

NaiaraLopezRojo/globalIRmap documentation built on Dec. 17, 2021, 5:19 a.m.