Plot Kolmogorov-Smirnov Curve

Generates a Kolmogorov-Smirnov (KS) plot from a dx object. The KS plot is a graphical tool used to assess the discriminatory power of a binary classification model by visualizing the difference between the cumulative distribution functions of the true positive rate (TPR) and false positive rate (FPR) across different thresholds.

Usage

dx_plot_ks(dx_obj)

Arguments

dx_obj

A dx object containing diagnostic measurements, including a rank data frame with columns for cumulative true positive rate (TPR) and false positive rate (FPR), as well as the predicted probabilities associated with each instance.

Value

A ggplot object representing the KS plot, which can be further customized as needed.

Details

The KS plot displays two lines representing the cumulative TPR and FPR, with the x-axis showing the prediction scores in descending order and the y-axis showing the cumulative proportion of positive and negative instances. The point where the difference between the TPR and FPR is maximal indicates the threshold with the highest potential for separating positive from negative instances. This point is marked with a dashed vertical line on the plot, and the KS statistic, representing the maximum distance between the two lines, is annotated on the plot. A higher KS statistic indicates a model with better discriminatory ability. The ideal model would have a KS plot with the TPR line close to the top-left corner and the FPR line close to the bottom-right corner, maximizing the distance between the two lines.

Examples

dx_obj <- dx(
  data = dx_heart_failure,
  true_varname = "truth",
  pred_varname = "predicted",
  outcome_label = "Heart Attack",
  setthreshold = .3
)
dx_plot_ks(dx_obj)