kingscross-hospitality-ai

Kings Cross Seasonal Demand Intelligence is an experimental urban analytics project focused on detecting and explaining demand volatility in high-footfall districts, using Kings Cross and Coal Drops Yard in London as a validation area. It ingests only public, explainable signals—weather, TfL transport status, local events from Eventbrite, venue density and proximity from Google Places, and time-of-day/seasonal context—and fuses them into a district-level busyness signal compared against seasonal baselines. The system classifies deviations using a fixed anomaly taxonomy covering demand anomalies (unexpected_peak, suppressed_demand, prolonged_peak, volatile_demand), timing anomalies (shifted_peak, missing_peak), and signal mismatch anomalies (transport_demand_mismatch, weather_demand_mismatch, event_demand_mismatch). Each anomaly includes severity, confidence based on signal agreement, persistence, a human-readable explanation, and contributing drivers. Outputs include JSON files for a dashboard, short-term forecast, demand history, classified anomalies, raw observations, and aggregated seasonal insights. The project is currently in a paused data-collection phase from late December 2025 to mid-January 2026, with model and anomaly logic frozen to capture clean seasonal signals without confounding code changes. The project explicitly avoids personal data, payment data, device tracking, and private venue data. It is positioned for property and asset managers, urban operators, place-making teams, PropTech partners, and consultants analysing seasonal or event-driven demand risk, with an architecture designed for replication across districts and cities.