barplot_var_stability function

Barplot variable stability

Produces a ggplot2 plot of stability (as SEM) of variable importance across models trained and tested across outer CV folds. Optionally overlays directionality for binary response or regression outcomes.

barplot_var_stability(
  x,
  final = TRUE,
  top = NULL,
  direction = 0,
  dir_labels = NULL,
  scheme = c("royalblue", "red"),
  breaks = NULL,
  percent = TRUE,
  level = 1,
  sort = TRUE
)

Arguments

• x: a nestcv.glmnet or nestcv.train fitted object
• final: Logical whether to restrict variables to only those which ended up in the final fitted model or to include all variables selected across all outer folds.
• top: Limits number of variables plotted. Set to NULL to plot all variables.
• direction: Integer controlling plotting of directionality for binary or regression models. 0 means no directionality is shown, 1 means directionality is overlaid as a colour, 2 means directionality is reflected in the sign of variable importance. Not available for multiclass caret models.
• dir_labels: Character vector for controlling the legend when direction = 1
• scheme: Vector of 2 colours for directionality when direction = 1
• breaks: Vector of continuous breaks for legend colour/size
• percent: Logical for nestcv.glmnet objects only, whether to scale coefficients to percentage of the largest coefficient in each model. If set to FALSE, model coefficients are shown and direction is ignored.
• level: For multinomial nestcv.glmnet models only, either an integer specifying which level of outcome is being examined, or the level can be specified as a character value.
• sort: Logical whether to sort by mean variable importance. Passed to var_stability().

Returns

A ggplot2 plot

See Also

var_stability()