Generates the data needed for the calibration plot — get_calibration_plot_data

Generates the data needed for the calibration plot. The calibration plot needs to separate the model predictions by risk groups. First the function separates the predictions in n_groups then computes the mean value of the model predictions and also the observed value. The observed value is the estimated value at the study time estimated using the proportion of events in each group.

Usage

get_calibration_plot_data_prop(
  model,
  data,
  n_groups,
  type = "predictions_aggregated"
)

Arguments

model: Model generated with mv_model_cox() or mv_model_logreg(). Needs the predictions parameter of the model, to generate it the function calculate_predictions() must be executed over the model. If we want to obtain also the recalibrated data the model must be initalize the recalibrated predictions with calculate_predictions_recalibrated_type_1() and calculate_predictions_recalibrated_type_2().
data: Data for what the observed predictions will be calculated.
n_groups: Number of groups that must be calculated.
type: Type of the predictions that the calibration plot data should be generated from: "predictions_aggregated", "predictions_recal_type_1" or "predictions_recal_type_2"

Value

tibble with the data ready to generate a calibration plot.

group	prediction	observed
1	0.03	0.05
...	...	...
n_group	0.84	0.79

Examples

if (FALSE) { # \dontrun{
model |>
  get_calibration_plot_data_surv(data = test_data, n_groups = 10, type = "predictions_aggregated")
} # }