Evidence boundary: Operational records and outcomes are synthetic. Public context establishes property fit and guest-facing value, not internal policy, costs, or performance.
The comp recommendation model consumes a curated recovery-case mart, not clean source rows.
raw_pms_reservations
+ raw_guest_profiles_crm
+ raw_service_tickets
+ raw_comp_ledger
+ raw_pos_outlet_charges
+ raw_reviews_surveys
+ raw_ops_daily
+ public_pricing_context
+ public_property_context
+ review_risk_context
+ local_demand_context
-> identity and reservation matching
-> dirty issue and comp taxonomy normalization
-> public quoted-rate, property, review, and local-demand context joins
-> source-quality flags and match confidence
-> data/marts/recovery_case_mart.csv
-> comp recommendation modelMart Quality Signals
- Recovery cases:
430 - Low reservation-match confidence cases:
10 - Cases with inferred severity:
60 - Cases with no historical comp record:
175
These issues are intentionally retained because the business problem is partly a data-wrangling problem.