MLwrap (version 0.3.0)

plot_calibration_curve: Plotting Calibration Curve

Description

The plot_calibration_curve() function generates calibration plots for binary classification models evaluating the agreement between predicted probabilities and observed class frequencies in binned prediction intervals. Implements reliability diagrams comparing empirical success rates within each probability bin against the predicted probability levels, identifying systematic calibration errors including overconfidence (predicted probabilities exceed observed frequencies) and underconfidence patterns across prediction ranges.

Usage

plot_calibration_curve(analysis_object)

Value

analysis_object

Arguments

analysis_object

Fitted analysis_object with 'fine_tuning()'.

Examples

# Note: For obtaining the calibration curve plot the user needs to
# complete till fine_tuning( ) function of the MLwrap pipeline and
# only with binary outcome.

wrap_object <- preprocessing(df = sim_data[1:300 ,],
                             formula = psych_well_bin ~ depression + resilience,
                             task = "classification")
wrap_object <- build_model(wrap_object, "Random Forest",
                           hyperparameters = list(mtry = 2, trees = 5))
set.seed(123) # For reproducibility
wrap_object <- fine_tuning(wrap_object, "Grid Search CV")

# And then, you can obtain the calibration curve plot.

plot_calibration_curve(wrap_object)