1. ECVP v2 execution purpose
1. Propósito de ejecución del ECVP v2
Apply the universal Module II protocol to the frozen Module I object and determine the evidence-capability families required for the selected Road Accident System variables and relationship architectures.
Aplicar el protocolo universal del Módulo II al objeto fijado del Módulo I y determinar las familias de capacidades evidenciales requeridas para las variables seleccionadas y arquitecturas de relación del Sistema de Accidentes de Tránsito.
Boundary: this document does not select Pearson, Cramér’s V, Mutual Information, Random Forest, SHAP, Moran’s I, or any other algorithm. It produces the scientific specification that Module III will use to select them.
Frontera: este documento no selecciona Pearson, V de Cramér, Mutual Information, Random Forest, SHAP, Moran’s I ni ningún otro algoritmo. Produce la especificación científica que utilizará el Módulo III para seleccionarlos.
2. Module I object inherited without modification
2. Objeto del Módulo I heredado sin modificación
| Inherited elementElemento heredado | Frozen case definitionDefinición fijada del caso |
|---|---|
| Scientific purposePropósito científico | Understand which recorded event, environmental, infrastructure, vehicle, temporal, and spatial conditions are related to accident severity.Comprender qué condiciones registradas del evento, ambientales, de infraestructura, vehículo, tiempo y espacio están relacionadas con la severidad del accidente. |
| Unit of analysisUnidad de análisis | One recorded road accident event.Un evento de accidente de tránsito registrado. |
| Analytical sampleMuestra analítica | 4,999 recorded accident events.4,999 eventos de accidentes registrados. |
| Primary objective Y₁Objetivo primario Y₁ | Accident_Severity — Slight → Serious → Fatal. |
| Secondary objective Y₂Objetivo secundario Y₂ | Number_of_Casualties — separate count consequence model. |
| Intermediate objective / event stateObjetivo intermedio / estado del evento | Number_of_Vehicles — event-complexity descriptor with simultaneous-state restrictions. |
| Scope restrictionRestricción de alcance | Severity conditional on an accident already being recorded; not accident-occurrence risk in the complete road-using population.Severidad condicionada a que el accidente ya haya sido registrado; no riesgo de ocurrencia en toda la población vial. |
3. Explicit validated variable architecture
3. Arquitectura explícita de variables validadas
3.1 Primary target
3.1 Objetivo primario
Y₁ = Accident_Severity
3.2 Exclusive severity drivers
3.2 Impulsores exclusivos de severidad
| Variable | System dimensionDimensión del sistema | Validated roleRol validado |
|---|---|---|
| Light_Conditions | Environment / visibilityAmbiente / visibilidad | Core explanatory variableVariable explicativa principal |
| Weather_Conditions | EnvironmentAmbiente | Core contextual explanatory variableVariable explicativa contextual principal |
| Road_Surface_Conditions | Environment / infrastructure stateAmbiente / estado de infraestructura | Physical road-state explanatory variableVariable explicativa del estado físico de la vía |
| Road_Type | InfrastructureInfraestructura | Road-configuration explanatory variableVariable explicativa de configuración vial |
| Vehicle_Type | VehicleVehículo | Vehicle vulnerability/class explanatory variableVariable explicativa de clase/vulnerabilidad vehicular |
| Junction_Detail | Infrastructure / interaction pointInfraestructura / punto de interacción | Junction-configuration explanatory variableVariable explicativa de configuración de intersección |
3.3 Shared temporal and spatial drivers
3.3 Impulsores temporales y espaciales compartidos
| Variable | System dimensionDimensión del sistema | Validated roleRol validado |
|---|---|---|
| Day_of_Week | TemporalTemporal | Weekly-cycle explanatory variableVariable explicativa de ciclo semanal |
| Time | TemporalTemporal | Hour-of-day and interaction variableVariable de hora del día e interacción |
| Latitude | SpatialEspacial | Coordinate component; valid only with LongitudeComponente de coordenada; válido solo con Longitude |
| Longitude | SpatialEspacial | Coordinate component; valid only with LatitudeComponente de coordenada; válido solo con Latitude |
| Local_Authority_(District) | Spatial / administrative contextContexto espacial / administrativo | Territorial segmentation variableVariable de segmentación territorial |
3.4 Conditional and supporting variables
3.4 Variables condicionales y de apoyo
| Variable | Validated statusEstado validado | RestrictionRestricción |
|---|---|---|
| Number_of_Casualties | Secondary objectiveObjetivo secundario | Separate consequence model; do not merge automatically with severity.Modelo separado de consecuencias; no mezclar automáticamente con severidad. |
| Number_of_Vehicles | Conditionally selected event-complexity variableVariable de complejidad del evento seleccionada condicionalmente | May be simultaneous with severity; avoid causal overstatement.Puede ser simultánea con severidad; evitar exageración causal. |
| Carriageway_Hazards | Conditionally selected contextual hazardPeligro contextual seleccionado condicionalmente | Requires frequency and missingness review.Requiere revisión de frecuencia y valores faltantes. |
| Junction_Control | Conditionally restrictedRestringida condicionalmente | Requires recoding and data-quality validation.Requiere recodificación y validación de calidad. |
| Police_Force | Administrative context onlySolo contexto administrativo | Source/jurisdiction control; not an intrinsic physical cause.Control de fuente/jurisdicción; no causa física intrínseca. |
3.5 Excluded or restricted variables
3.5 Variables excluidas o restringidas
| Variable | DecisionDecisión | ReasonRazón |
|---|---|---|
| Accident_Index | Excluded from evidence testingExcluida de pruebas evidenciales | Traceability identifier without explanatory meaning.Identificador de trazabilidad sin significado explicativo. |
| Urban_or_Rural_Area | Not applicable for comparisonNo aplicable para comparación | No useful variation in the sample.Sin variación útil en la muestra. |
| Speed_limit | Restricted / descriptive onlyRestringida / solo descriptiva | Extremely limited variation makes comparative evidence fragile.Variación extremadamente limitada vuelve frágil la evidencia comparativa. |
4. Relationship configuration inventory
4. Inventario de configuraciones de relación
| ID | Explicit relationshipRelación explícita | PurposePropósito |
|---|---|---|
| REL-001 | Light_Conditions ↔ Accident_Severity | Evaluate severity differences across visibility states.Evaluar diferencias de severidad entre estados de visibilidad. |
| REL-002 | Weather_Conditions ↔ Accident_Severity | Evaluate environmental association, dependence, and information contribution.Evaluar asociación ambiental, dependencia y aporte informacional. |
| REL-003 | Road_Surface_Conditions ↔ Accident_Severity | Evaluate physical-road-state evidence.Evaluar evidencia del estado físico de la vía. |
| REL-004 | Road_Type ↔ Accident_Severity | Evaluate road-configuration evidence.Evaluar evidencia de configuración vial. |
| REL-005 | Vehicle_Type ↔ Accident_Severity | Evaluate severity profile by vehicle class.Evaluar perfil de severidad por clase vehicular. |
| REL-006 | Junction_Detail ↔ Accident_Severity | Evaluate conflict-point configuration evidence.Evaluar evidencia de configuración del punto de conflicto. |
| REL-007 | Day_of_Week ↔ Accident_Severity | Evaluate weekly pattern differences.Evaluar diferencias de patrón semanal. |
| REL-008 | Time ↔ Accident_Severity | Evaluate hour-of-day and temporal-state evidence.Evaluar evidencia por hora del día y estado temporal. |
| REL-009 | {Latitude, Longitude} ↔ Accident_Severity | Evaluate local concentration and geographic heterogeneity.Evaluar concentración local y heterogeneidad geográfica. |
| REL-010 | Local_Authority_(District) ↔ Accident_Severity | Evaluate territorial segmentation and context.Evaluar segmentación territorial y contexto. |
| REL-011 | {Weather_Conditions, Road_Surface_Conditions, Light_Conditions} ↔ Accident_Severity | Evaluate grouped environmental dependence, redundancy, synergy, and prediction.Evaluar dependencia ambiental agrupada, redundancia, sinergia y predicción. |
| REL-012 | {Road_Type, Junction_Detail, Vehicle_Type} ↔ Accident_Severity | Evaluate infrastructure–vehicle joint structure.Evaluar estructura conjunta infraestructura–vehículo. |
| REL-013 | {Day_of_Week, Time, Light_Conditions} ↔ Accident_Severity | Evaluate temporal-visibility configuration.Evaluar configuración temporal–visibilidad. |
| REL-014 | {Latitude, Longitude, Local_Authority_(District)} ↔ Accident_Severity | Evaluate coordinate-level and administrative spatial structure.Evaluar estructura espacial por coordenadas y territorio administrativo. |
| REL-015 | Number_of_Vehicles ↔ Accident_Severity | Conditional event-complexity relationship; simultaneous-state caution.Relación condicional de complejidad del evento; cautela por simultaneidad. |
| REL-016 | Accident_Severity ↔ Number_of_Casualties | Cross-objective consequence architecture; no automatic direction or causality.Arquitectura de consecuencias entre objetivos; sin dirección o causalidad automática. |
5. Master Evidence Capability Matrix
5. Matriz Maestra de Capacidades Evidenciales
Legend: R = Required · C = Conditional · D = Deferred · N = Not Required · IA = Insufficient Structure / Not Applicable.
Leyenda: R = Requerida · C = Condicional · D = Diferida · N = No requerida · IA = Estructura insuficiente / No aplicable.
| Explicit variable / groupVariable / grupo explícito | Association | Dependence | Information | Time | Space | Prediction | Explanation | Causality | Integration |
|---|---|---|---|---|---|---|---|---|---|
| Light_Conditions | R | R | R | C | N | R | C | D | R |
| Weather_Conditions | R | R | R | C | C | R | C | D | R |
| Road_Surface_Conditions | R | R | R | C | C | R | C | D | R |
| Road_Type | R | R | R | N | C | R | C | D | R |
| Vehicle_Type | R | R | R | N | C | R | C | D | R |
| Junction_Detail | R | R | R | C | C | R | C | D | R |
| Day_of_Week | R | R | R | R | N | R | C | D | R |
| Time | R | R | R | R | N | R | C | D | R |
| {Latitude, Longitude} | R | R | C | N | R | C | C | D | R |
| Local_Authority_(District) | R | R | R | N | R | R | C | D | R |
| {Weather_Conditions, Road_Surface_Conditions, Light_Conditions} | R | R | R | C | C | R | R | D | R |
| {Road_Type, Junction_Detail, Vehicle_Type} | R | R | R | N | C | R | R | D | R |
| {Day_of_Week, Time, Light_Conditions} | R | R | R | R | N | R | R | D | R |
| {Latitude, Longitude, Local_Authority_(District)} | R | R | C | N | R | R | R | D | R |
| Number_of_Vehicles | C | C | C | C | C | C | C | D | C |
| Carriageway_Hazards | C | C | C | C | C | C | C | D | C |
| Junction_Control | IA | IA | IA | IA | IA | IA | IA | D | N |
| Police_Force | C | C | N | N | C | C | C | D | C |
| Accident_Severity ↔ Number_of_Casualties | R | R | R | C | C | C | C | D | R |
6. Capability Justification Matrix
6. Matriz de Justificación de Capacidades
| CapabilityCapacidad | StatusEstado | Explicit variables / groupsVariables / grupos explícitos | Scientific justificationJustificación científica | Critical limitLímite crítico |
|---|---|---|---|---|
| Association | Required | Light_Conditions; Weather_Conditions; Road_Surface_Conditions; Road_Type; Vehicle_Type; Junction_Detail; Day_of_Week; Time; Latitude/Longitude; Local_Authority_(District); validated groups | The case must characterize whether severity distributions differ across categorical, temporal, spatial, and grouped system states.El caso debe caracterizar si las distribuciones de severidad difieren entre estados categóricos, temporales, espaciales y agrupados del sistema. | Association does not prove dependence completeness, prediction, or causality.Asociación no demuestra dependencia completa, predicción ni causalidad. |
| Dependence | Required | All core variables and priority groups | Mixed categorical and multivariate structures may contain nonlinear, conditional, segmented, and emergent dependence not captured by one association form.Las estructuras mixtas categóricas y multivariadas pueden contener dependencia no lineal, condicional, segmentada y emergente no capturada por una sola forma asociativa. | Dependence remains observational and may arise from mixture or omitted variables.La dependencia sigue siendo observacional y puede surgir de mezcla o variables omitidas. |
| Information | Required | All core variables; environmental group; infrastructure–vehicle group; temporal–visibility group | The case must quantify uncertainty reduction, rank unique contribution, detect redundancy, and identify synergy among grouped variables.El caso debe cuantificar reducción de incertidumbre, ordenar contribución única, detectar redundancia e identificar sinergia entre variables agrupadas. | Information estimates require adequate sample support and careful treatment of sparse categories.Las estimaciones informacionales requieren soporte muestral adecuado y tratamiento cuidadoso de categorías escasas. |
| Time | Required / Conditional | Day_of_Week; Time; Light_Conditions; Weather_Conditions; Road_Surface_Conditions; Junction_Detail; Number_of_Vehicles conditional | Weekly cycle, hour-of-day, visibility state, and possible periodic or lagged configurations may affect severity evidence.El ciclo semanal, hora del día, estado de visibilidad y posibles configuraciones periódicas o rezagadas pueden afectar la evidencia de severidad. | The dataset must contain valid event date/order for longitudinal claims; hour and weekday alone support segmentation but not full time-series causality.El dataset debe contener fecha/orden del evento válidos para afirmaciones longitudinales; hora y día solamente sustentan segmentación, no causalidad completa de series. |
| Space | Required / Conditional | Latitude; Longitude; Local_Authority_(District); Weather_Conditions; Road_Surface_Conditions; Road_Type; Vehicle_Type; Junction_Detail | The case contains coordinate and administrative structures capable of supporting concentration, neighborhood, local heterogeneity, and territorial comparison.El caso contiene estructuras de coordenadas y administrativas capaces de sustentar concentración, vecindad, heterogeneidad local y comparación territorial. | Counts of recorded accidents do not represent population exposure or occurrence risk without denominators.Los conteos de accidentes registrados no representan exposición poblacional ni riesgo de ocurrencia sin denominadores. |
| Prediction | Required | Core explanatory variables and validated grouped predictor sets | The case purpose includes evaluating whether selected conditions can anticipate unseen Accident_Severity beyond a defensible baseline.El propósito incluye evaluar si las condiciones seleccionadas pueden anticipar Accident_Severity no visto más allá de una línea base defendible. | Prediction requires leakage-safe validation and class-sensitive evaluation; in-sample fit is insufficient.Predicción requiere validación sin fuga y evaluación sensible a clases; ajuste dentro de muestra es insuficiente. |
| Explanation | Conditional → Required after predictive model approval | Exact variables retained by the validated predictive architecture | Stakeholders will need global, local, interaction, and objective-specific interpretation of any approved model.Los interesados necesitarán interpretación global, local, interactiva y específica por objetivo de cualquier modelo aprobado. | Model explanation cannot be presented as the true mechanism of accident severity.Explicación del modelo no puede presentarse como mecanismo verdadero de severidad. |
| Causality | Deferred | All candidate drivers; no validated treatment variable exists | The current dataset is observational and defines no treatment, intervention assignment, counterfactual comparison, or identification design.El dataset actual es observacional y no define tratamiento, asignación de intervención, comparación contrafactual ni diseño de identificación. | No causal algorithm may be selected until a separate causal architecture is validated.No puede seleccionarse algoritmo causal hasta validar una arquitectura causal separada. |
| Integration | Required | All approved relationship units and both active objectives | Association, dependence, information, temporal, spatial, predictive, and explanatory results must be reconciled without treating them as equivalent.Resultados asociativos, de dependencia, información, tiempo, espacio, predicción y explicación deben reconciliarse sin tratarlos como equivalentes. | Strong evidence for Accident_Severity cannot be transferred silently to Number_of_Casualties.Evidencia fuerte para Accident_Severity no puede transferirse silenciosamente a Number_of_Casualties. |
7. Time and Space eligibility gates
7. Puertas de elegibilidad de Tiempo y Espacio
Time gate — Conditionally passed
Puerta temporal — Aprobada condicionalmente
Day_of_Week and Time support cyclic and intraday segmentation. Full temporal dependence, lag, trend, and regime analysis require a valid event-date sequence and sufficient repeated periods.
Day_of_Week y Time sustentan segmentación cíclica e intradía. Dependencia temporal completa, rezagos, tendencias y regímenes requieren secuencia válida de fechas del evento y periodos repetidos suficientes.
Space gate — Passed for recorded-event geography
Puerta espacial — Aprobada para geografía de eventos registrados
Latitude and Longitude form the required coordinate pair; Local_Authority_(District) supports territorial context. Analysis is limited to geographic structure among recorded accidents, not population-normalized accident risk.
Latitude y Longitude forman el par de coordenadas requerido; Local_Authority_(District) sustenta contexto territorial. El análisis se limita a estructura geográfica entre accidentes registrados, no riesgo normalizado por población.
8. Prediction, Explanation, and Causality gates
8. Puertas de Predicción, Explicación y Causalidad
| CapabilityCapacidad | Case verdictVeredicto del caso | Required conditionCondición requerida |
|---|---|---|
| Prediction | Required and eligible | Ordinal target preserved; leakage-safe split; rare Fatal class protected; baseline and objective-specific metrics defined.Objetivo ordinal preservado; partición sin fuga; clase Fatal rara protegida; línea base y métricas específicas definidas. |
| Explanation | Conditional until model validation, then Required | Only explain a model that has passed predictive validation; retain global/local distinction and correlated-variable warnings.Explicar solamente un modelo que haya pasado validación predictiva; conservar distinción global/local y advertencias por variables correlacionadas. |
| Causality | Deferred | Requires a separate treatment, comparison, population, timing, causal graph, and identification design.Requiere tratamiento, comparación, población, tiempo, grafo causal y diseño de identificación separados. |
9. Evidence Integration Requirements
9. Requisitos de Integración de Evidencia
Preserve native resultsPreservar resultados nativosDo not collapse association, information, prediction, and spatial evidence into one undocumented number.No colapsar asociación, información, predicción y evidencia espacial en un número no documentado.
Separate objectivesSeparar objetivosMaintain independent evidence specifications for Accident_Severity and Number_of_Casualties.Mantener especificaciones independientes para Accident_Severity y Number_of_Casualties.
Preserve contradictionsPreservar contradiccionesStrong association with weak prediction, or global evidence with weak local evidence, must remain visible.Asociación fuerte con predicción débil, o evidencia global con evidencia local débil, debe permanecer visible.
No double countingSin doble conteoMethods using the same variables and sample do not automatically create independent confirmation.Métodos que usan las mismas variables y muestra no crean automáticamente confirmación independiente.
TraceabilityTrazabilidadEvery result must trace to relationship ID, variables, capability, future method, assumptions, and version.Cada resultado debe rastrearse a ID de relación, variables, capacidad, método futuro, supuestos y versión.
10. Evidence Specification Summary (ESS)
10. Evidence Specification Summary (ESS)
This ESS is the official Module II output and formal Module III input object for Case 01.
Este ESS es la salida oficial del Módulo II y el objeto formal de entrada al Módulo III para el Caso 01.
10.1 ESS — Case identity and objectives
10.1 ESS — Identidad del caso y objetivos
| ESS_ID | KESM-ESS-M2-C01-v2 |
| Case_ID | Module_2_Method_Validation_Case_01_Road_Accident_System |
| Module I reference | Module_1_Method_Validation_Case_01_Road_Accident_System_v3 |
| Primary objective Y₁ | Accident_Severity |
| Secondary objective Y₂ | Number_of_Casualties — separate consequence model |
| Intermediate state | Number_of_Vehicles — conditional event-complexity role |
10.2 ESS — Variables by validated group
10.2 ESS — Variables por grupo validado
| ID | GroupGrupo | Explicit variablesVariables explícitas | Primary objectiveObjetivo principal |
|---|---|---|---|
| X¹ | Environmental / visibilityAmbiental / visibilidad | Light_Conditions; Weather_Conditions; Road_Surface_Conditions | Accident_Severity |
| X² | Infrastructure / interaction pointInfraestructura / punto de interacción | Road_Type; Junction_Detail; Junction_Control (conditional) | Accident_Severity |
| X³ | Vehicle / event complexityVehículo / complejidad del evento | Vehicle_Type; Number_of_Vehicles (conditional) | Accident_Severity |
| X⁴ | TemporalTemporal | Day_of_Week; Time | Accident_Severity |
| X⁵ | SpatialEspacial | Latitude; Longitude; Local_Authority_(District) | Accident_Severity |
| X⁶ | Conditional contextual hazardsPeligros contextuales condicionales | Carriageway_Hazards | Accident_Severity |
| X⁷ | Administrative contextContexto administrativo | Police_Force | Control onlySolo control |
10.3 ESS — Variables by evidence family
10.3 ESS — Variables por familia evidencial
| Evidence familyFamilia evidencial | StatusEstado | Explicit variables / groupsVariables / grupos explícitos |
|---|---|---|
| Association | Required | Light_Conditions; Weather_Conditions; Road_Surface_Conditions; Road_Type; Vehicle_Type; Junction_Detail; Day_of_Week; Time; Latitude; Longitude; Local_Authority_(District); X¹; X²; X³; X⁴; X⁵ |
| Dependence | Required | All core variables and groups X¹–X⁵; Number_of_Vehicles and Carriageway_Hazards conditional |
| Information | Required | All core variables; X¹ environmental; X²/X³ infrastructure–vehicle; X⁴ temporal; cross-objective Accident_Severity ↔ Number_of_Casualties |
| Time | Required / Conditional | Day_of_Week; Time; Light_Conditions; Weather_Conditions; Road_Surface_Conditions; Junction_Detail; Number_of_Vehicles conditional |
| Space | Required / Conditional | Latitude; Longitude; Local_Authority_(District); Weather_Conditions; Road_Surface_Conditions; Road_Type; Vehicle_Type; Junction_Detail |
| Prediction | Required | X¹ Environmental; X² Infrastructure; X³ Vehicle; X⁴ Temporal; X⁵ Spatial; conditional X⁶; target Accident_Severity |
| Explanation | Conditional → Required after model validation | Exact variables retained by the approved predictive model |
| Causality | Deferred | No validated treatment variable; all apparent drivers remain observational candidates |
| Integration | Required | REL-001 through REL-016; Y₁ Accident_Severity; Y₂ Number_of_Casualties |
10.4 ESS — Critical assumptions and constraints
10.4 ESS — Supuestos y restricciones críticas
- Accident_Severity is ordinal and must not be treated as an arbitrary numeric distance without explicit justification.
- Accident_Severity es ordinal y no debe tratarse como distancia numérica arbitraria sin justificación explícita.
- Fatal outcomes are comparatively rare and require class-sensitive evaluation.
- Los resultados fatales son comparativamente raros y requieren evaluación sensible a clases.
- Latitude and Longitude must remain a pair.
- Latitude y Longitude deben permanecer como par.
- Number_of_Vehicles may be simultaneous with accident consequences and cannot be interpreted causally by default.
- Number_of_Vehicles puede ser simultánea con las consecuencias y no puede interpretarse causalmente por defecto.
- Junction_Control cannot advance until recoding and quality validation are completed.
- Junction_Control no puede avanzar hasta completar recodificación y validación de calidad.
- Spatial results describe recorded-event geography, not exposure-normalized accident risk.
- Los resultados espaciales describen geografía de eventos registrados, no riesgo normalizado por exposición.
- Causality remains deferred.
- Causalidad permanece diferida.
10.5 ESS executive specification
10.5 Especificación ejecutiva del ESS
======================================================== EVIDENCE SPECIFICATION SUMMARY — CASE 01 ======================================================== ESS_ID: KESM-ESS-M2-C01-v2 PRIMARY OBJECTIVE: Y₁ = Accident_Severity SECONDARY OBJECTIVE: Y₂ = Number_of_Casualties Separate consequence model EXPLICIT CORE VARIABLES: Light_Conditions Weather_Conditions Road_Surface_Conditions Road_Type Vehicle_Type Junction_Detail Day_of_Week Time Latitude Longitude Local_Authority_(District) CONDITIONAL VARIABLES: Number_of_Vehicles Carriageway_Hazards Junction_Control Police_Force EXCLUDED / RESTRICTED: Accident_Index Urban_or_Rural_Area Speed_limit EVIDENCE FAMILIES: Association = REQUIRED Dependence = REQUIRED Information = REQUIRED Time = REQUIRED / CONDITIONAL Space = REQUIRED / CONDITIONAL Prediction = REQUIRED Explanation = CONDITIONAL → REQUIRED AFTER MODEL VALIDATION Causality = DEFERRED Integration = REQUIRED MODULE III READINESS: CONDITIONALLY VALIDATED CONDITIONS BEFORE METHOD SELECTION: 1. Preserve ordinal target structure. 2. Validate event-date structure for full temporal methods. 3. Preserve Latitude + Longitude as a pair. 4. Validate sparse categories and missingness. 5. Recode and validate Junction_Control before inclusion. 6. Define leakage-safe predictive validation. 7. Keep causal methods prohibited until a causal design exists. ========================================================
11. Formal Module III handoff contract
11. Contrato formal de entrega al Módulo III
Module III SHALL
El Módulo III DEBERÁ
Select and compare candidate methods for each Required or eligible Conditional evidence family using the explicit variables, groups, relationships, objectives, assumptions, and constraints in this ESS.
Seleccionar y comparar métodos candidatos para cada familia Requerida o Condicional elegible utilizando variables, grupos, relaciones, objetivos, supuestos y restricciones explícitas de este ESS.
Module III SHALL NOT
El Módulo III NO DEBERÁ
Redefine Accident_Severity, replace selected variables, regroup them silently, include excluded fields, activate Junction_Control without validation, or select causal methods.
Redefinir Accident_Severity, sustituir variables seleccionadas, reagruparlas silenciosamente, incluir campos excluidos, activar Junction_Control sin validación ni seleccionar métodos causales.
Return rule: any missing variable, invalid role, insufficient temporal/spatial structure, or unsupported relationship must be returned to Module I or Module II. Module III cannot repair the scientific architecture silently.
Regla de retorno: toda variable faltante, rol inválido, estructura temporal/espacial insuficiente o relación no sustentada debe regresar al Módulo I o II. El Módulo III no puede reparar silenciosamente la arquitectura científica.
12. Module II validation verdict
12. Veredicto de validación del Módulo II
CONDITIONALLY VALIDATED FOR MODULE III
VALIDADO CONDICIONALMENTE PARA EL MÓDULO III
The Evidence Specification Summary is complete, variable-explicit, objective-specific, and scientifically justified. Module III may begin method selection for Association, Dependence, Information, Prediction, eligible Time and Space structures, and future Explanation. Causality remains formally deferred.
El Evidence Specification Summary está completo, contiene variables explícitas, es específico por objetivo y está científicamente justificado. El Módulo III puede comenzar selección de métodos para Asociación, Dependencia, Información, Predicción, estructuras elegibles de Tiempo y Espacio y futura Explicación. Causalidad permanece formalmente diferida.
The case no longer asks “Which algorithm do we know?” It asks “Which algorithm best satisfies each validated evidence specification?”
El caso ya no pregunta “¿Qué algoritmo conocemos?”. Pregunta “¿Qué algoritmo satisface mejor cada especificación evidencial validada?”.