The Observation Theory EncyclopediaFrom TSKAboutBy kindBy chapterBy Lean fileLedgerProvenance

outlier concept

DefinitionA row the reader cannot place. The statistical detector calls a row an outlier when its distance from the mean in units of spread passes a threshold, and no more than one over the threshold squared of the weight can. Chapter 10. Also z-score.
ExampleA row 3.6 spreads from the mean is an outlier at threshold 3, and by Chebyshev at most one ninth of the weight lies past 3 spreads.
BookData Mining as Observation, draft 0.2, commit f3914f0; entry id outlier, kind concept.
Statusno ledger row names this entry. Corrections: none recorded.
Defining equationnone
Assumptions and scope
  • A row the reader cannot place. The statistical detector scores a row by its distance from the mean in units of spread, which is zero at the mean, unchanged when data, mean, and spread are rescaled and shifted together, and past a threshold t exactly when the row lies more than t spreads from the mean. By Chebyshev the weight of rows beyond t is at most the mean squared score over t squared.
  • The four detector families read different coordinates, distance from the mean, distance to neighbours, density relative to neighbours, and cluster membership, and the observer’s outliers, the anti-hubs, are the rows no detector of the data alone can see.
Prior artnone recorded
Evidencelean/DataMiningAsObservation/Outlier.lean, lean/DataMiningAsObservation/Mahalanobis.lean
Reviewednot yet reviewed; generated 2026-09-10 from records at the commits on the provenance page.
outlieroutlier
A row the reader cannot place.

Equation

none

Conditions

Conditions are curated in entries.toml rather than read from a record.

Ledger

none

First stated

Chapter 0 section 0.16 and chapter 10 section 10.1 of Data Mining as Observation, after TSK chapter 9, with the hierarchical typing in turboquant-pro/turboquant_pro/anatomy.py:98-170.

Measurements

none

Failures and corrections

none

Invariance envelope

none declared

Machine checked

lean/DataMiningAsObservation/Outlier.lean, theorems zscore_mean, zscore_affine, outlier_iff, zscore_reflect, chebyshev, at observation-data-mining f3914f0; what the check covers is stated in the book’s appendix C.

lean/DataMiningAsObservation/Mahalanobis.lean, theorems dM2_nonneg, dM2_mean, dM2_scale, dM2_eq_whitened, dM2_antitone_in_variance, at observation-data-mining f3914f0; what the check covers is stated in the book’s appendix C.

Used in

Data Mining as Observation primer S, chapters 0, 2, 3, 10.

Related

Mahalanobis distance; anti-hub; abstention; whitening; density.

See also

Book equations stated beside the entry’s terms, not defining it: 0.33, 10.3, 0.6.

Ledger rows that cite the entry’s records without naming it: NEG-14.

Sources-table rows that share a record with the entry without naming it: chapter 10 section 10.1, chapter 10 section 10.3.

Status

Generated 2026-09-10 by encyclopedia/generate.py; book at observation-data-mining f3914f0; the commit of every record is listed in the encyclopedia’s provenance.

← outer productoutput metric →