what is roc

In data and stats, ROC usually means Receiver Operating Characteristic (and the ROC curve).

Quick meaning

• ROC is a way to judge how good a yes/no (binary) classifier or diagnostic test is at separating two classes, like “diseased vs healthy” or “spam vs not spam.”

• The ROC curve plots:
  • True Positive Rate (sensitivity) on the y‑axis
  • False Positive Rate (1 − specificity) on the x‑axis
    for many different decision thresholds.

In simple terms, you slide the threshold from very strict to very lenient, and the ROC curve shows how the trade‑off between catching more positives (sensitivity) and making more false alarms (false positives) changes.

Why people care about ROC

• It helps compare different models or tests independently of any one chosen threshold.

• The Area Under the Curve (AUC) is a single number summary: the closer AUC is to 1, the better the model can distinguish between the two classes; an AUC around 0.5 is basically random guessing.

In practice, ROC/AUC is standard in:

• Medical diagnostics (e.g., blood test to detect a disease).

• Machine learning classification (spam filters, fraud detection, etc.).

Quick example

Imagine a model gives each person a risk score for a disease between 0 and 1.

• If you label “diseased” when score ≥ 0.9, you get few false positives but miss many true cases.
• If you label “diseased” when score ≥ 0.2, you catch almost everyone with the disease but flag many healthy people.

Every such cutoff gives a pair (FPR, TPR); plotting all these pairs yields the ROC curve.

Other meaning: Rate of Change (ROC)

In finance/trading, ROC often means Rate of Change , a momentum indicator measuring how fast price is changing over a look‑back period.

That ROC is unrelated to the ROC curve in statistics, but both share the idea of comparing change or performance across a range.