Loading...
Loading...
Plot ROC curves, calculate AUC, and find the optimal classification threshold. Drag the slider to see how metrics change.
A Receiver Operating Characteristic curve plots True Positive Rate (y-axis) vs False Positive Rate (x-axis) at every classification threshold. A perfect classifier hugs the top-left corner. A random one follows the diagonal.
Area Under the Curve summarizes ROC performance in one number (0 to 1). AUC = 1.0 means perfect separation. AUC = 0.5 means no better than coin flip. Example: a spam filter with AUC = 0.95 correctly ranks 95% of spam above non-spam.
Paste your labels and predicted probabilities above, or load a sample dataset.
Supports comma, space, or newline-separated values.
An ROC curve (Receiver Operating Characteristic) plots the trade-off between True Positive Rate (sensitivity) and False Positive Rate (1 − specificity) at every classification threshold.
Ranges from 0 to 1. A random classifier scores 0.5. Higher AUC means better class separation. It's threshold-independent - useful for comparing models.
Found using Youden's J statistic (max TPR − FPR). Maximizes the distance from the random diagonal. For imbalanced datasets, consider precision-recall curves instead.
Best for balanced datasets. Compares classifiers regardless of threshold. Use in medical screening, fraud detection, and any binary classification task.
Can be misleading with highly imbalanced classes (e.g., 99% negative). Use Precision-Recall curves when the positive class is rare.
AUC (Area Under the ROC Curve) measures how well a binary classifier distinguishes between positive and negative classes across all thresholds. An AUC of 1.0 means perfect classification, while 0.5 means the model performs no better than random guessing.
This calculator uses Youden's J statistic, which maximizes the difference between True Positive Rate and False Positive Rate (TPR - FPR). The optimal threshold is shown as the green dot on the ROC curve.
Use ROC AUC when your dataset has roughly balanced classes. For imbalanced datasets where the positive class is rare (like fraud detection with 0.1% fraud), precision-recall curves give a more informative picture of model performance.
True Positive Rate (TPR, or recall) is the proportion of actual positives correctly identified. False Positive Rate (FPR) is the proportion of actual negatives incorrectly classified as positive. The ROC curve plots TPR against FPR at every threshold.
AUC above 0.9 indicates excellent discrimination. Between 0.8 and 0.9 is good. Between 0.7 and 0.8 is fair and may need improvement. Below 0.7 suggests the model struggles to separate the classes effectively.