# Schema definitions for energy test data datasets

schemas:
  electricity_prices:
    columns:
      - name: "timestamp"
        type: "datetime64[ns]"
        description: "Timestamp of price observation"
      - name: "region"
        type: "category"
        description: "Market region code"
      - name: "day_ahead_price"
        type: "float32"
        unit: "EUR/MWh"
        description: "Day-ahead market clearing price"
      - name: "real_time_price"
        type: "float32"
        unit: "EUR/MWh"
        description: "Real-time market price"
      - name: "capacity_price"
        type: "float32"
        unit: "EUR/MWh"
        description: "Capacity market price"
      - name: "regulation_price"
        type: "float32"
        unit: "EUR/MWh"
        description: "Frequency regulation price"
      - name: "volume_mw"
        type: "float32"
        unit: "MW"
        description: "Traded volume"

  battery_capacity:
    columns:
      - name: "timestamp"
        type: "datetime64[ns]"
        description: "Timestamp of battery state"
      - name: "battery_id"
        type: "category"
        description: "Unique battery identifier"
      - name: "capacity_mwh"
        type: "float32"
        unit: "MWh"
        description: "Total storage capacity"
      - name: "charge_level_mwh"
        type: "float32"
        unit: "MWh"
        description: "Current energy stored"
      - name: "charge_rate_mw"
        type: "float32"
        unit: "MW"
        description: "Current charging rate (positive) or discharging (negative)"
      - name: "discharge_rate_mw"
        type: "float32"
        unit: "MW"
        description: "Maximum discharge rate"
      - name: "efficiency"
        type: "float32"
        description: "Round-trip efficiency (0-1)"

  renewable_generation:
    columns:
      - name: "timestamp"
        type: "datetime64[ns]"
        description: "Timestamp of generation measurement"
      - name: "source"
        type: "category"
        description: "Renewable source type (solar, wind, hydro)"
      - name: "plant_id"
        type: "category"
        description: "Unique plant identifier"
      - name: "generation_mw"
        type: "float32"
        unit: "MW"
        description: "Actual generation output"
      - name: "forecast_mw"
        type: "float32"
        unit: "MW"
        description: "Forecasted generation"
      - name: "actual_mw"
        type: "float32"
        unit: "MW"
        description: "Actual measured generation (after correction)"
      - name: "capacity_factor"
        type: "float32"
        description: "Capacity utilization factor (0-1)"

  conventional_generation:
    columns:
      - name: "timestamp"
        type: "datetime64[ns]"
        description: "Timestamp of generation measurement"
      - name: "plant_id"
        type: "category"
        description: "Unique plant identifier"
      - name: "fuel_type"
        type: "category"
        description: "Primary fuel type (gas, coal, nuclear)"
      - name: "generation_mw"
        type: "float32"
        unit: "MW"
        description: "Current generation output"
      - name: "marginal_cost"
        type: "float32"
        unit: "EUR/MWh"
        description: "Short-run marginal cost"
      - name: "heat_rate"
        type: "float32"
        unit: "MMBtu/MWh"
        description: "Thermal efficiency metric"

  load_profiles:
    columns:
      - name: "timestamp"
        type: "datetime64[ns]"
        description: "Timestamp of load measurement"
      - name: "region"
        type: "category"
        description: "Region code"
      - name: "load_mw"
        type: "float32"
        unit: "MW"
        description: "Actual system load"
      - name: "forecast_mw"
        type: "float32"
        unit: "MW"
        description: "Load forecast"
      - name: "weather_temp"
        type: "float32"
        unit: "Celsius"
        description: "Average temperature"
      - name: "humidity"
        type: "float32"
        unit: "%"
        description: "Relative humidity"

  data_centers:
    columns:
      - name: "timestamp"
        type: "datetime64[ns]"
        description: "Timestamp of demand measurement"
      - name: "data_center_id"
        type: "category"
        description: "Data center identifier"
      - name: "location"
        type: "category"
        description: "Geographic location"
      - name: "power_demand_mw"
        type: "float32"
        unit: "MW"
        description: "Current power demand"
      - name: "max_bid_price"
        type: "float32"
        unit: "EUR/MWh"
        description: "Maximum price willing to pay"
      - name: "client_type"
        type: "category"
        description: "Client type (bitcoin, enterprise, etc.)"

  bitcoin_mining:
    columns:
      - name: "timestamp"
        type: "datetime64[ns]"
        description: "Timestamp of mining measurement"
      - name: "pool_id"
        type: "category"
        description: "Mining pool identifier"
      - name: "hashrate_ths"
        type: "float32"
        unit: "TH/s"
        description: "Mining pool hashrate"
      - name: "btc_price_eur"
        type: "float32"
        unit: "EUR"
        description: "Bitcoin price in EUR"
      - name: "power_efficiency_th_per_mw"
        type: "float32"
        unit: "TH/s per MW"
        description: "Mining efficiency"
      - name: "power_demand_mw"
        type: "float32"
        unit: "MW"
        description: "Power consumption for mining"
      - name: "revenue_eur_per_mwh"
        type: "float32"
        unit: "EUR/MWh"
        description: "Mining revenue per MWh of electricity"
      - name: "profit_eur_per_mwh"
        type: "float32"
        unit: "EUR/MWh"
        description: "Mining profit after electricity cost"
      - name: "electricity_cost"
        type: "float32"
        unit: "EUR/MWh"
        description: "Electricity cost for mining"

  transmission_capacity:
    columns:
      - name: "source_region"
        type: "category"
        description: "Source region code"
      - name: "target_region"
        type: "category"
        description: "Target region code"
      - name: "capacity_mw"
        type: "float32"
        unit: "MW"
        description: "Maximum transmission capacity"
      - name: "direction"
        type: "category"
        description: "Transmission direction"
      - name: "efficiency"
        type: "float32"
        description: "Transmission efficiency (0-1)"

  transmission_cost:
    columns:
      - name: "source_region"
        type: "category"
        description: "Source region code"
      - name: "target_region"
        type: "category"
        description: "Target region code"
      - name: "cost_eur_mwh"
        type: "float32"
        unit: "EUR/MWh"
        description: "Total transmission cost per MWh"
      - name: "loss_percent"
        type: "float32"
        unit: "%"
        description: "Transmission loss percentage"
      - name: "congestion_surcharge_eur_mwh"
        type: "float32"
        unit: "EUR/MWh"
        description: "Additional congestion charge"
      - name: "fee_eur_mwh"
        type: "float32"
        unit: "EUR/MWh"
        description: "Transmission fee"

validation_rules:
  electricity_prices:
    - column: "day_ahead_price"
      min: -500
      max: 3000
    - column: "real_time_price"
      min: -500
      max: 5000

  battery_capacity:
    - column: "charge_level_mwh"
      min: 0
      check_max: "capacity_mwh"
    - column: "efficiency"
      min: 0.5
      max: 1.0

  renewable_generation:
    - column: "generation_mw"
      min: 0
    - column: "capacity_factor"
      min: 0
      max: 1.0

  conventional_generation:
    - column: "generation_mw"
      min: 0
    - column: "heat_rate"
      min: 5
      max: 15

  load_profiles:
    - column: "load_mw"
      min: 0
    - column: "weather_temp"
      min: -30
      max: 50

  data_centers:
    - column: "power_demand_mw"
      min: 0
    - column: "max_bid_price"
      min: 0

  bitcoin_mining:
    - column: "hashrate_ths"
      min: 0
      max: 1000000
    - column: "btc_price_eur"
      min: 1000
      max: 200000
    - column: "power_efficiency_th_per_mw"
      min: 50
      max: 150
    - column: "power_demand_mw"
      min: 10
      max: 1000
    - column: "revenue_eur_per_mwh"
      min: 0
      max: 500

  transmission_capacity:
    - column: "capacity_mw"
      min: 100
      max: 10000
    - column: "efficiency"
      min: 0.9
      max: 1.0

  transmission_cost:
    - column: "cost_eur_mwh"
      min: 0
      max: 50
    - column: "loss_percent"
      min: 0
      max: 15