ROC: Receiver Operating Characteristic (ROC)

In fake: Flexible Data Simulation Using the Multivariate Normal Distribution

Receiver Operating Characteristic (ROC)

Description

Computes the True and False Positive Rates (TPR and FPR, respectively) and Area Under the Curve (AUC) by comparing the true (observed) and predicted status using a range of thresholds on the predicted score.

Usage

ROC(observed, predicted, n_thr = NULL)

Arguments

observed	vector of binary outcomes.
predicted	vector of predicted scores.
n_thr	number of thresholds to use to construct the ROC curve. For faster computations on large data, values below length(predicted)-1 can be used.

Value

A list with:

TPR	True Positive Rate.
FPR	False Positive Rate.
AUC	Area Under the Curve.

See Also

Other goodness of fit functions: Concordance()

Examples

# Data simulation
set.seed(1)
simul <- SimulateRegression(
  n = 500, pk = 20,
  family = "binomial", ev_xy = 0.8
)

# Logistic regression
fitted <- glm(simul$ydata ~ simul$xdata, family = "binomial")$fitted.values

# Constructing the ROC curve
roc <- ROC(predicted = fitted, observed = simul$ydata)
plot(roc)

fake documentation built on April 14, 2023, 12:37 a.m.