ROC curve

From WikiMD.org
Jump to navigation Jump to search

ROC Curve

The ROC curve (pronounced /ɑːr oʊ siː kɜːrv/), or Receiver Operating Characteristic curve, is a graphical plot that illustrates the diagnostic ability of a binary classifier system as its discrimination threshold is varied. The ROC curve is created by plotting the true positive rate (TPR) against the false positive rate (FPR) at various threshold settings.

Etymology

The term "ROC" originated from signal detection theory developed during World War II for the analysis of radar signals. The name "Receiver Operating Characteristic" was coined as it reflected the ability of the radar receiver operator to correctly identify enemy aircraft.

Related Terms

  • True Positive Rate: Also known as sensitivity, it measures the proportion of actual positives that are correctly identified as such.
  • False Positive Rate: Also known as fall-out, it measures the proportion of actual negatives that are incorrectly identified as positives.
  • Binary Classifier: A type of classification model that categorizes data into one of two groups.
  • Discrimination Threshold: The point of separation where the probabilities of belonging to either of the two groups are equal.

Usage

The ROC curve is widely used in medicine, radiology, biometrics and various fields of machine learning. It is a popular tool for predictive models that need to balance sensitivity and specificity.

Calculation

The ROC curve is generated by plotting the cumulative distribution function (area under the probability distribution from minus infinity to the discrimination threshold) of the true positive rate, against the cumulative distribution function of the false positive rate, at various threshold settings.

Advantages

The ROC curve is a comprehensive and flexible tool. It allows the user to select the optimal model based on their specific cost/benefit tradeoff. It also allows for the comparison of different models.

Disadvantages

The ROC curve does not provide a single metric of performance, which can make model selection difficult. It also assumes that the costs of false positives and false negatives are roughly equal, which may not be the case in all applications.

External links

Esculaap.svg

This WikiMD article is a stub. You can help make it a full article.


Languages: - East Asian 中文, 日本, 한국어, South Asian हिन्दी, Urdu, বাংলা, తెలుగు, தமிழ், ಕನ್ನಡ,
Southeast Asian Indonesian, Vietnamese, Thai, မြန်မာဘာသာ, European español, Deutsch, français, русский, português do Brasil, Italian, polski