Elamur Groupantskaggle-text-command-group-prediction
The groupant beside tools/kaggle/src/main.rs in the Mova repository: what the code computes, what it never does, how to go back from its outputs, its invariants, cost and checks.
Facts
Implementation material. A Mova code model describes this code at one source revision: partial models, finite checks and their stated scope. It is not a reviewed general exposition, and its labels are shown as written, never raised.
Components
| component | label | where |
|---|---|---|
| approximant | partial | approximant |
| ast | partial | ../syntax.json |
| cost | symbolic source derived | invariant-and-cost |
| descriptant | present | descriptant |
| invariant | partial | invariant-and-cost |
| tests | executed bounded checks | tests |
Obernants
The typed relations around the forward model: the behaviour complement (what the code never does), the reverse reconstruction (every input compatible with an output) and specialised views.
| id | type | label | where |
|---|---|---|---|
kaggle-text-command-group-prediction-behavior_complement | behavior_complement | partial source derived model | behavior_complement |
kaggle-text-command-group-prediction-reverse_reconstruction | reverse_reconstruction | partial source derived model | reverse_reconstruction |
kaggle-text-command-group-prediction-key_union | key_union | partial source derived model | key_union |
kaggle-text-command-group-prediction-known_projection | known_projection | partial source derived model | known_projection |
kaggle-text-command-group-prediction-override_precedence | override_precedence | partial source derived model | override_precedence |
kaggle-text-command-group-prediction-missing_label_neutrality | missing_label_neutrality | partial source derived model | missing_label_neutrality |
Model text
descriptant
kaggle-text-command-group-prediction
Epoch 2026-10-31
Full source read preceded changes. Prospective model predicted branch/order, resource and information-loss behavior. Tests exercise actual helpers on bounded synthetic inputs; command workflows remain unexecuted. Production source prefix is preserved; only inline tests added.
Build two string-key buckets per person, dedup unequal-id neighbor entries; known id map from training; count known female/Master mates' Some(true)/Some(false). Female death override then Master survival override then gender fallback.
Executable percentage approximation is unsupported: categorical parsing/effects have no useful certified metric here. Partial syntax excludes macro expansion and name resolution.
approximant
Forward relation
Build two string-key buckets per person, dedup unequal-id neighbor entries; known id map from training; count known female/Master mates' Some(true)/Some(false). Female death override then Master survival override then gender fallback.
Domain
Bounded N; unique ids for simple graph semantics; fare string rounding and exact ticket equality.
models
Typed source models
All models are partial source-derived relations at this manifest snapshot. Scope is one call/trace under the stated domain; no universal semantic proof.
behavior_complement
No edge may arise solely through a third passenger; all unknown or conflicting mates cannot trigger either unanimous override.
Domain: Bounded N; unique ids for simple graph semantics; fare string rounding and exact ticket equality. Observation: No self-id edges; duplicate records can overwrite known id. Labels None count neither lived nor died; direct mates only. Assumptions: bounded resources and source branch semantics. Oracle: selected actual-helper tests in tests.md constrain only their finite fixtures; remaining branches are source-derived.
reverse_reconstruction
Neighbor relation admits alternative ticket and surname/fare bucket decompositions; WCG boolean admits many compatible labels, not unique train.
Domain: Bounded N; unique ids for simple graph semantics; fare string rounding and exact ticket equality. Observation: No self-id edges; duplicate records can overwrite known id. Labels None count neither lived nor died; direct mates only. Assumptions: bounded resources and source branch semantics. Oracle: selected actual-helper tests in tests.md constrain only their finite fixtures; remaining branches are source-derived.
key_union
G(a)=union of same n:surname:fare and t:ticket ids minus a; insertion duplicate suppression preserves one neighbor id.
Domain: Bounded N; unique ids for simple graph semantics; fare string rounding and exact ticket equality. Observation: No self-id edges; duplicate records can overwrite known id. Labels None count neither lived nor died; direct mates only. Assumptions: bounded resources and source branch semantics. Oracle: selected actual-helper tests in tests.md constrain only their finite fixtures; remaining branches are source-derived.
known_projection
Known is train collected by id, later duplicate wins; mates absent from Known are discarded before woman/child filter.
Domain: Bounded N; unique ids for simple graph semantics; fare string rounding and exact ticket equality. Observation: No self-id edges; duplicate records can overwrite known id. Labels None count neither lived nor died; direct mates only. Assumptions: bounded resources and source branch semantics. Oracle: selected actual-helper tests in tests.md constrain only their finite fixtures; remaining branches are source-derived.
override_precedence
Female and Master simultaneously: all-dead branch wins before survival branch; mixed lived/dead takes gender fallback.
Domain: Bounded N; unique ids for simple graph semantics; fare string rounding and exact ticket equality. Observation: No self-id edges; duplicate records can overwrite known id. Labels None count neither lived nor died; direct mates only. Assumptions: bounded resources and source branch semantics. Oracle: selected actual-helper tests in tests.md constrain only their finite fixtures; remaining branches are source-derived.
missing_label_neutrality
Some(true) increments lived; Some(false) died; None neither, so 'all' wording means observed labeled mates, not every mate.
Domain: Bounded N; unique ids for simple graph semantics; fare string rounding and exact ticket equality. Observation: No self-id edges; duplicate records can overwrite known id. Labels None count neither lived nor died; direct mates only. Assumptions: bounded resources and source branch semantics. Oracle: selected actual-helper tests in tests.md constrain only their finite fixtures; remaining branches are source-derived.
tests
Evidence
Selected exact tests: ["wave33::features_groups_and_empty_learning"].
Each runs twice in debug and twice in release with normalized named outcomes compared. Compiled production-boundary mutants are recorded in the external wave33 text receipts with original/restored hashes and positive reruns. Existing negative assertions are separate from compiled mutants. No CLI, download, teacher, corpus or model execution. For this fragment, source-derived paths not reached by these tests remain unverified.
invariant-and-cost
Invariant and cost
Invariant
No self-id edges; duplicate records can overwrite known id. Labels None count neither lived nor died; direct mates only.
Cost
Bucket materialization O(N log N); dense bucket dedup O(N^3) worst case, neighbor memory O(N^2); each prediction rebuilds.
Arithmetic and allocation premises are conditional. No timing or percentage surrogate claimed.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/tools/kaggle/src/main.rs.groupant/group-prediction.groupant/