Visualizer for loss functions

Visualize one or multiple loss functions.

Super class

vistool::Visualizer -> VisualizerLossFuns

Public fields

losses

(list)
List of LossFunction objects.

task_type

(character(1))
Task type (regr or classif).

y_pred

(numeric())
Predicted values.

y_true

(numeric())
True values.

input_type

(character(1))
Input scale for classification tasks: "score" (margin‑based, default) or "probability".

Methods

Public methods

• VisualizerLossFuns$new()

• VisualizerLossFuns$plot()

• VisualizerLossFuns$add_points()

• VisualizerLossFuns$clone()

Inherited methods

• vistool::Visualizer$add_annotation()
• vistool::Visualizer$add_contours()
• vistool::Visualizer$resolve_layer_colors()
• vistool::Visualizer$save()
• vistool::Visualizer$set_theme()
• vistool::Visualizer$theme()

---

Method new()

Creates a new instance of this R6 class.

Usage

VisualizerLossFuns$new(
  losses,
  y_pred = NULL,
  y_true = NULL,
  input_type = "auto",
  n_points = 1000L,
  ...
)

Arguments

losses

(list)
List of LossFunction objects.

y_pred

(numeric())
Predicted values. Optional.

y_true

(numeric())
True values. Optional.

input_type

(character(1))
Desired input scale. One of "auto", "score", "probability". "auto" (default) chooses the common input_default of the supplied losses.

n_points

(integer(1)) Default resolution used when calling $plot() without specifying n_points. Defaults to 1000.

...

Additional arguments (currently unused).

---

Method plot()

Create and return the ggplot2 plot with model-specific layers.

Usage

VisualizerLossFuns$plot(
  n_points = NULL,
  y_curves = "both",
  line_width = NULL,
  line_col = NULL,
  line_type = NULL,
  ...
)

Arguments

n_points

(integer(1))
Number of points to use for plotting the loss functions. Defaults to the value configured when constructing the visualizer (via as_visualizer()), falling back to 1000 if not set.

y_curves

(character(1))
When input_type = "probability", choose which curves to display: "both", "y1", or "y0". Default is "both".

line_width

(numeric())
Line widths for different loss functions. If NULL, uses default width of 1.2 for all lines.

line_col

(character())
Line colors for different loss functions. If NULL, uses automatic color assignment.

line_type

(character())
Line types for different loss functions. If NULL, uses "solid" for all lines.

...

Additional arguments passed to the parent plot method.

Returns

A ggplot2 object.

---

Method add_points()

Add points to the loss function visualization with optional vertical lines. Points are automatically positioned on the loss curve by evaluating the selected loss function at the given x-coordinates.

Usage

VisualizerLossFuns$add_points(
  x,
  loss_id = NULL,
  show_line = TRUE,
  color = "auto",
  size = NULL,
  alpha = NULL,
  line_color = NULL,
  line_alpha = NULL
)

Arguments

x

(numeric())
x-coordinates where to place points on the loss curves. These represent the residual values (y-f for regression, y*f for classification scores, or probabilities for classification probability mode).

loss_id

(character(1))
ID of the loss function to use for y-value calculation. If NULL (default), uses the first loss function.

show_line

(logical(1))
If TRUE (default), draws vertical lines from points to x-axis.

color

(character(1))
Color of the points and lines. Use "auto" for automatic color assignment. Default is "auto".

size

(numeric(1))
Size of the points. If NULL, uses theme$point_size. Default is NULL.

alpha

(numeric(1))
Alpha transparency of points and lines. If NULL, uses theme$alpha. Default is NULL.

line_color

(character(1))
Color of vertical lines. If NULL, uses the same color as points.

line_alpha

(numeric(1))
Alpha transparency of vertical lines. If NULL, uses alpha * 0.7.

---

Method clone()

The objects of this class are cloneable with this method.

Usage

VisualizerLossFuns$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.