Skip to content
Last Dictionary
Plane
Machine LearningEvaluation & Benchmarks

specificity

nounAlso: true negative rate, TNR, selectivity

In plain wordsOf all the cases that really were negative, the share a test correctly cleared; a highly specific test rarely raises false alarms on healthy people.

  1. The fraction of actual negatives that a classifier or test correctly identifies as negative: TN/(TN + FP), also called the true negative rate. It is the counterpart of recallrecall, n.: Of all the items that really were positive, the share the model managed to find; a cancer screen with high recall misses very few people who actually have cancer. (sensitivity), which measures the fraction of positives caught, and the sensitivity–specificity pair is the standard way to report diagnostic tests in medicine. Specificity equals 1 minus the false positive rate, so the ROC curveROC curve, n.: A graph showing how a detector's hit rate and false-alarm rate change together as it is made more or less trigger-happy; the closer the curve hugs the top-left corner, the better. plots sensitivity against 1 − specificity across decision thresholds. Like sensitivity, it does not depend on the prevalence of positives, whereas precisionprecision, n.: Of all the items a model flagged as positive, the share that really were positive; a spam filter with high precision rarely sends genuine emails to the spam folder. (positive predictive value) does; consequently a test with high sensitivity and specificity can still produce mostly false alarms when screening for a rare condition. In machine learning, specificity is reported mainly in medical and security applications; elsewhere precision and recall are more usual.

    On the external hospital's data the screening model reached 92% sensitivity at 95% specificity.

Specificity=TNTN+FP\mathrm{Specificity} = \frac{TN}{TN + FP}

Origin

Introduced together with sensitivity by the biostatistician Jacob Yerushalmy in a 1947 study of observer variation in X-ray diagnosis of tuberculosis, and standardized in clinical epidemiology for reporting diagnostic test performance.

Related

See also
  • recallrecall, n.: Of all the items that really were positive, the share the model managed to find; a cancer screen with high recall misses very few people who actually have cancer.
  • false positivefalse positive, n.: A false alarm: a test or AI system raises a flag, for example for illness, fraud or danger, when nothing of the kind is really there.
  • ROC curveROC curve, n.: A graph showing how a detector's hit rate and false-alarm rate change together as it is made more or less trigger-happy; the closer the curve hugs the top-left corner, the better.
  • confusion matrixconfusion matrix, n.: A table that counts a model's right and wrong answers by type, showing for each true category how often each category was predicted, so you can see exactly which things it mixes up.
  • precisionprecision, n.: Of all the items a model flagged as positive, the share that really were positive; a spam filter with high precision rarely sends genuine emails to the spam folder.
  • AUCAUC, abbr.: A single number from 0 to 1 summarizing how well a model ranks real positive cases above negative ones: 0.5 is no better than a coin flip and 1.0 is perfect.
Contrast with
  • precisionprecision, n.: Of all the items a model flagged as positive, the share that really were positive; a spam filter with high precision rarely sends genuine emails to the spam folder.

Sources

  1. Yerushalmy, J. (1947). Statistical Problems in Assessing Methods of Medical Diagnosis, with Special Reference to X-Ray Techniques. Public Health Reports 62(40).

Mentioned in

false positivefalse positive, n.: A false alarm: a test or AI system raises a flag, for example for illness, fraud or danger, when nothing of the kind is really there., AUCAUC, abbr.: A single number from 0 to 1 summarizing how well a model ranks real positive cases above negative ones: 0.5 is no better than a coin flip and 1.0 is perfect., confusion matrixconfusion matrix, n.: A table that counts a model's right and wrong answers by type, showing for each true category how often each category was predicted, so you can see exactly which things it mixes up., decision thresholddecision threshold, n.: The cut-off score above which a model's prediction counts as 'yes'; moving it trades missed cases against false alarms, like setting how sensitive a smoke alarm should be., false negativefalse negative, n.: A miss: a test or AI system gives the all-clear, for example no illness, no fraud or no danger, when the problem is in fact there., recallrecall, n.: Of all the items that really were positive, the share the model managed to find; a cancer screen with high recall misses very few people who actually have cancer., ROC curveROC curve, n.: A graph showing how a detector's hit rate and false-alarm rate change together as it is made more or less trigger-happy; the closer the curve hugs the top-left corner, the better.