🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
184 lines
4.5 KiB
Markdown
184 lines
4.5 KiB
Markdown
---
|
|
id: "US-ML-001"
|
|
title: "Valuacion automatica de propiedad"
|
|
type: "User Story"
|
|
epic: "IAI-008"
|
|
status: "Draft"
|
|
story_points: 13
|
|
priority: "Alta"
|
|
sprint: "-"
|
|
project: "inmobiliaria-analytics"
|
|
created_date: "2026-01-04"
|
|
updated_date: "2026-01-04"
|
|
---
|
|
|
|
# US-ML-001: Valuacion automatica de propiedad
|
|
|
|
---
|
|
|
|
## User Story
|
|
|
|
**Como** agente inmobiliario
|
|
**Quiero** obtener una valuacion automatica de una propiedad
|
|
**Para** proporcionar a mis clientes una estimacion de precio basada en datos
|
|
|
|
---
|
|
|
|
## Descripcion
|
|
|
|
Implementar el modelo AVM (Automated Valuation Model) que estime el valor de mercado de propiedades basandose en caracteristicas, ubicacion y condiciones de mercado. El modelo debe ser preciso (MAPE < 10%) y proporcionar explicaciones de los factores que influyen en la valuacion.
|
|
|
|
---
|
|
|
|
## Criterios de Aceptacion
|
|
|
|
### Funcionales
|
|
|
|
- [ ] Usuario ingresa caracteristicas de propiedad
|
|
- [ ] Sistema retorna valor estimado con rango
|
|
- [ ] Sistema muestra score de confianza
|
|
- [ ] Sistema muestra precio por m2
|
|
- [ ] Sistema lista propiedades comparables
|
|
|
|
### Tecnicos
|
|
|
|
- [ ] MAPE < 10% en test set
|
|
- [ ] Latencia < 500ms
|
|
- [ ] Modelo versionado en MLflow
|
|
- [ ] API disponible en `/api/v1/ml/valuation/predict`
|
|
|
|
---
|
|
|
|
## Mockup de Interfaz
|
|
|
|
```
|
|
+------------------------------------------+
|
|
| VALUACION AUTOMATICA |
|
|
+------------------------------------------+
|
|
| |
|
|
| Tipo: [Casa v] Transaccion: [Venta v] |
|
|
| |
|
|
| Recamaras: [3] Banos: [2.5] |
|
|
| Superficie construida: [180] m2 |
|
|
| Superficie terreno: [250] m2 |
|
|
| Antiguedad: [5] anos |
|
|
| |
|
|
| Direccion: [Av. Americas 1234_______] |
|
|
| Colonia: [Providencia] |
|
|
| Ciudad: [Guadalajara] |
|
|
| |
|
|
| [ CALCULAR VALUACION ] |
|
|
| |
|
|
+------------------------------------------+
|
|
| RESULTADO |
|
|
+------------------------------------------+
|
|
| |
|
|
| Valor Estimado: $4,850,000 MXN |
|
|
| Rango: $4,600,000 - $5,100,000 |
|
|
| Precio/m2: $26,944 |
|
|
| Confianza: 87% |
|
|
| |
|
|
| Factores principales: |
|
|
| + Ubicacion premium (+12%) |
|
|
| + Superficie generosa (+8%) |
|
|
| - Antiguedad moderada (-3%) |
|
|
| |
|
|
| Comparables cercanos: |
|
|
| | Direccion | Precio | m2 | $/m2 | |
|
|
| |-----------|--------|-----|------| |
|
|
| | Americas | 4.9M | 190 | 25.8K| |
|
|
| | Lopez M. | 4.7M | 175 | 26.9K| |
|
|
| |
|
|
+------------------------------------------+
|
|
```
|
|
|
|
---
|
|
|
|
## Tareas Tecnicas
|
|
|
|
| # | Tarea | Estimacion |
|
|
|---|-------|------------|
|
|
| 1 | Feature engineering pipeline | 8h |
|
|
| 2 | Entrenar modelo XGBoost | 4h |
|
|
| 3 | Entrenar modelo LightGBM | 4h |
|
|
| 4 | Implementar ensemble | 4h |
|
|
| 5 | API endpoint FastAPI | 4h |
|
|
| 6 | Integracion frontend | 8h |
|
|
| 7 | Tests unitarios y de integracion | 6h |
|
|
| 8 | Documentacion y model card | 4h |
|
|
|
|
**Total estimado:** 42h (~5 dias)
|
|
|
|
---
|
|
|
|
## Datos de Entrada
|
|
|
|
```yaml
|
|
input:
|
|
required:
|
|
- property_type: string
|
|
- transaction_type: string
|
|
- construction_m2: number
|
|
- latitude: number
|
|
- longitude: number
|
|
|
|
optional:
|
|
- land_m2: number
|
|
- bedrooms: integer
|
|
- bathrooms: number
|
|
- parking_spaces: integer
|
|
- age_years: integer
|
|
- postal_code: string
|
|
- amenities: string[]
|
|
```
|
|
|
|
---
|
|
|
|
## Datos de Salida
|
|
|
|
```yaml
|
|
output:
|
|
estimated_price: number
|
|
price_range:
|
|
min: number
|
|
max: number
|
|
price_per_m2: number
|
|
confidence_score: number # 0-1
|
|
factors:
|
|
- name: string
|
|
impact: string # "positive" | "negative"
|
|
magnitude: number # %
|
|
comparables:
|
|
- id: string
|
|
address: string
|
|
price: number
|
|
surface_m2: number
|
|
price_per_m2: number
|
|
similarity_score: number
|
|
```
|
|
|
|
---
|
|
|
|
## Definition of Done
|
|
|
|
- [ ] MAPE < 10% validado
|
|
- [ ] API responde en < 500ms
|
|
- [ ] UI implementada y funcional
|
|
- [ ] Comparables se muestran correctamente
|
|
- [ ] Tests pasan (unit + integration)
|
|
- [ ] Model card documentado
|
|
- [ ] Deployed a staging
|
|
|
|
---
|
|
|
|
## Dependencias
|
|
|
|
- IA-007 completado (datos de propiedades)
|
|
- Modelo de datos de propiedades
|
|
- Infraestructura ML (FastAPI, MLflow)
|
|
|
|
---
|
|
|
|
**Asignado a:** -
|
|
**Sprint:** -
|