Threshold and Cutpoint Table¶
Reference artifact — instantiates Boundary-Sensitive Segmentation Design
Records the exact cutpoints that assign each case to a segment, together with the inclusivity, rounding, and missing-value conventions that make the assignment reproducible.
The Threshold and Cutpoint Table is the plain, auditable artifact that pins down where the cuts fall and how a borderline placement resolves — the deterministic lookup that turns a raw value into a segment. Its distinguishing idea is that reproducibility lives in the conventions, not just the numbers: two analysts hand the same case only agree if the table also fixes which side of a cut owns the exact boundary value, how raw readings are rounded before lookup, and where missing or out-of-range values go. It carries no theory of why the cuts sit where they do and assigns no meaning to the resulting segments — it is the mechanical rule stripped to something a spreadsheet can execute the same way every time.
Example¶
A city's air-quality dashboard converts a continuous PM2.5 concentration into the U.S. EPA Air Quality Index categories. The published AQI breakpoint table does the work: a 24-hour concentration is truncated to one decimal place before lookup, and each category owns a closed concentration range. A sensor reads 35.46 µg/m³ — right at the Moderate / Unhealthy-for-Sensitive-Groups seam near ≈35.4 µg/m³. The table's conventions decide the outcome deterministically: truncation drops it to 35.4, and the inclusivity rule keeps 35.4 in the lower category, so the reading posts as Moderate rather than tipping the whole city into a health advisory. Change only the rounding rule — round instead of truncate — and the identical air now reads Unhealthy for Sensitive Groups. The table makes that consequential choice explicit and repeatable instead of leaving it to whichever script ran the lookup.
How it works¶
What distinguishes the table from a bare list of numbers is the set of conventions it locks down alongside the cutpoints:
- Cutpoints — the ordered break values themselves.
- Inclusivity — for each cut, which side owns the exact boundary value (
≤vs<), so no case is unassigned or double-assigned. - Quantization — the rounding or truncation applied to a raw value before lookup, which silently manufactures or removes exact-boundary cases.
- Residual handling — where missing, negative, or out-of-range values land (own category, nearest bin, or reject).
The result is a static, human-readable, version-stampable rule. The table records cutpoints; it does not decide them — those arrive from an evidence basis or a data-derived derivation upstream.
Tuning parameters¶
- Inclusivity convention (
≤vs<) — which side of each cut claims the exact boundary value; flips a small but nonzero mass of cases and, at a consequential cut, real allocations. - Rounding vs truncation — how raw values are quantized before lookup; coarser quantization piles more cases exactly on the cutpoints.
- Residual defaults — the destination for missing and out-of-range values; a permissive default hides data-quality problems, a strict one rejects real cases.
- Open vs closed extremes — whether the top and bottom segments are unbounded or capped.
- Stated precision — how many significant figures the cutpoints carry, which can imply an accuracy the underlying evidence never had.
When it helps, and when it misleads¶
Its strength is total reproducibility: anyone can replay an assignment and get the same answer, and every borderline case has a pre-declared resolution. That auditability is exactly why it is trusted — and exactly why it is dangerous. The table launders arbitrary cutpoints into the same authoritative-looking artifact as well-justified ones; the format certifies nothing about the placement. Its structural hazard is the notch (cliff) effect — because treatment jumps discontinuously at a cut, a case an arbitrarily small distance across the line receives a categorically different outcome, which both distorts behavior near the seam and rewards gaming.[n1] The classic misuse is to choose the cutpoints to produce a desired allocation and then present the finished table as neutral machinery. The discipline that guards against this is to keep the table downstream of a real evidence basis and a sensitivity test, and to treat a heavy mass of exact-boundary cases as a signal that the cut, the rounding rule, or the whole crisp-cut framing needs rethinking.
How it implements the components¶
boundary_placement_rule— the table is the placement rule: the enumerated cutpoints plus the inclusivity, quantization, and residual conventions that make each assignment reproducible.edge_case_and_tie_break_policy— it resolves the deterministic slice of the edge problem: exact-boundary, rounded-to-the-cut, missing, and out-of-range cases each get a fixed, pre-declared destination.
It does not model the domain or justify the cuts (that evidence comes from a derivation such as Image-Region Segmentation Pipeline or a clustering workflow), assign positive meaning to the segments (Score-Banding Model), soften near-boundary cases into a graded band (Overlap-Band Assignment), or route genuine judgment calls to people — cases it cannot resolve deterministically escalate to the Manual Boundary Review Queue.
Related¶
- Instantiates: Boundary-Sensitive Segmentation Design — the table is the concrete, reproducible placement artifact the archetype's assignments run on.
- Consumes: Clustering-to-Boundary Workflow and Change-Point Segmentation supply the cutpoint values the table records.
- Sibling mechanisms: Score-Banding Model · Overlap-Band Assignment · Binning and Discretization Scheme · Manual Boundary Review Queue · Segmented Holdout Validation · Boundary Sensitivity Analysis · Boundary Change Log
Editorial Notes¶
Form Classification¶
Form family: Rule, Policy & Commitment
Rationale: Threshold And Cutpoint Table is defined in the frozen evidence as: Records the exact cutpoints that assign each case to a segment, together with the inclusivity, rounding, and missing-value conventions that make the assignment reproducible. Its operative deployed or enacted form is therefore Rule, Policy & Commitment.
Nearest alternative: Representation, Specification & Plan — Representation, Specification & Plan can support this mechanism, but the evidence centers the concrete operation described above rather than the alternative family's defining operation.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Threshold and cutpoint table derives most directly from statistics' measurement, sampling, inference, and experimental-design tradition; its defining operation is to records the exact cutpoints that assign each case to a segment, together with the inclusivity, rounding, and missing-value conventions that make the assignment reproducible.
Related originating lineages:
- Data Science & Analytics — Data science's telemetry, modeling, profiling, and monitoring tradition provides a formative adjacent lineage for the same threshold and cutpoint table operation.
- Mathematics — Mathematical modeling, proof, and abstract-structure practice supplies a parallel or contributing lineage for the mechanism's defining operation: records the exact cutpoints that assign each case to a segment, together with the inclusivity, rounding, and missing-value conventions that make the assignment reproducible.
Review resolution: Both blind reviewers independently select statistics_experimental_design as the primary historical origin for the concrete operation—Records the exact cutpoints that assign each case to a segment, together with the inclusivity, rounding, and missing-value conventions that make the assignment reproducible. The queued differences concern alternate origin disagreement, origin mode disagreement, domain reach disagreement, encyclopedia synthesis disagreement, not the primary lineage. I retain every alternate that either reviewer explains, without a numeric cap, and choose origin_mode=cross_disciplinary_synthesis because the reviewers' combined evidence identifies material construction from multiple disciplines. domain_reach=multi_domain records later portability rather than multiplying historical origins; confidence=high is the conservative shared evidentiary level, and encyclopedia_synthesis=true preserves either reviewer's affirmative synthesis finding.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
The same table can serve a scrupulously governed segmentation or a purely arbitrary one — it encodes placement, not legitimacy. That is why it is deliberately kept as a thin, versioned artifact downstream of evidence and revision controls: when the cutpoints change, the change belongs in a Boundary Change Log, not silently in the lookup.
[n1] The notch or cliff effect is the discontinuity a hard cutpoint creates, where crossing the line by an arbitrarily small amount changes obligation, benefit, or label discretely. It is a well-known hazard in tax and benefits design (the "benefit cliff"), and it is the reason a consequential cut should be sensitivity-tested rather than trusted because the table looks precise. ↩