# Campus Energy Intelligence: data card
600 independent fictional building-day scenarios. Seed 202601. No real campus measurements or personal records. These are independently sampled scenarios, not a temporal sequence; random stratified splitting is appropriate only for this classroom construction. Real repeated-building or forecasting studies need group/time evaluation.

Columns: scenario_id is an identifier (exclude); temperature_c, occupancy, operating_hours and area_m2 are inputs assumed available before prediction; energy_kwh is the measured end-of-day outcome; high_demand is 1 when energy_kwh >= 125; end_of_day_meter_kwh is only available after the day ends and is an intentional leakage trap. Never use the latter two outcomes as prediction inputs.

Four allowed inputs. Twenty-four missing values each in temperature and occupancy; imputation learns only from training. Energy equation: 15 + .24 occupancy + 3.1 hours + .045 area + 1.6 max(temperature - 25,0) + Gaussian noise SD9. All energy units kWh. Round inputs as stored. No deployment claim, causal interpretation or representative accuracy claim. Original synthetic data, reusable for learning with attribution to MindForgeAI Jigyasa.