The Observation Theory EncyclopediaFrom TSKAboutBy kindBy chapterBy Lean fileLedgerProvenance

baseline instrument

DefinitionThe simplest scorer a claim must beat under the same protocol, chance, a constant, an untrained encoder, or a single model. A score below it is a finding. Chapter 6 section 6.3 and chapter 7 section 7.4.
ExampleAn untrained encoder scored 0.50 on the rights corpus and the trained one 0.467, below the baseline.
BookData Mining as Observation, draft 0.2, commit f3914f0; entry id baseline, kind instrument.
Statusno ledger row names this entry. Corrections: none recorded.
Defining equationnone
Assumptions and scope
  • The simplest scorer a claim must beat, chance, a constant, an untrained encoder, or a single model, run under the same protocol. A baseline that passes the bar makes the bar vacuous, and a score below the baseline is a finding.
  • The rights encoder read 0.467, below the untrained baseline, and the formula’s comparison with the ensemble is fair only against the single-model baseline inside one fold protocol.
Prior artnone recorded
Evidencelean/DataMiningAsObservation/ChanceLevel.lean, lean/DataMiningAsObservation/Bar.lean
Reviewednot yet reviewed; generated 2026-09-10 from records at the commits on the provenance page.
00.51statisticbaselinesystembar
The simplest scorer the claim must beat, under the same protocol.

Equation

none

Conditions

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

Ledger

none

First stated

Chapter 6 section 6.3 and chapter 7 section 7.4 of Data Mining as Observation, with the single-model baseline in constraint-gap/review/INDETERMINATES.md:1-40.

Measurements

Where the book states it Numbers, as the book’s sources table records them Source
chapter 8 section 8.6 rights encoder 0.467 below untrained baseline, the AUROC lesson, cross-corpus same-sign fix xbse/README.md:160-172; xbse/experiments/rights_r6_summary.json

Failures and corrections

none

Invariance envelope

none declared

Machine checked

lean/DataMiningAsObservation/ChanceLevel.lean, theorems aurocNum_const, auroc_const, weight_const, tau_const, at observation-data-mining f3914f0; what the check covers is stated in the book’s appendix C.

lean/DataMiningAsObservation/Bar.lean, theorems passes_anti, passes_mono, discriminates_iff, no_bar_of_null_ge, exists_bar_of_lt, vacuous_of_null_passes, at observation-data-mining f3914f0; what the check covers is stated in the book’s appendix C.

Used in

Data Mining as Observation chapters 5, 6, 7, 8, 10, 11, 12, 13, 14.

Related

chance level; null model; bar; ensemble.

See also

Book equations stated beside the entry’s terms, not defining it: 8.2, 0.28, 6.2.

Ledger rows that cite the entry’s records without naming it: NEG-4, GO-B-legal (035→036).

Sources-table rows that share a record with the entry without naming it: chapter 6 section 6.3, chapter 7 section 7.3, chapter 7 section 7.4.

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.

← barbase rate →