Add transmission datasets and update mining data

Add two new static datasets for cross-region arbitrage calculations:
- transmission_capacity: region-to-region capacity limits (20 rows)
- transmission_cost: transmission costs per path (20 rows)

Update mining dataset with EUR pricing and power metrics:
- Change btc_price_usd to btc_price_eur
- Add power_efficiency_th_per_mw, power_demand_mw
- Add revenue_eur_per_mwh, profit_eur_per_mwh
- Remove mining_profitability column

Changes include:
- scripts/02_fetch_historical.py: rewrite fetch_bitcoin_mining_data()
- scripts/01_generate_synthetic.py: add transmission data generators
- config/data_config.yaml: add transmission config, update bitcoin config
- config/schema.yaml: add 2 new schemas, update bitcoin_mining schema
- scripts/03_process_merge.py: add 2 new datasets
- scripts/04_validate.py: add 2 new datasets
- test/test_data.py: update for new datasets and bitcoin price reference

Total datasets: 9 (734,491 rows, 17.89 MB)

config/data_config.yaml

@@ -93,4 +93,12 @@ data_center:
 
 bitcoin:
   hashrate_range: [150, 250]  # EH/s
-  mining_efficiency_range: [25, 35]  # J/TH
+  power_efficiency_range: [80, 120]  # TH/s per MW
+  eur_usd_rate: 0.92  # For converting to EUR base price
+
+transmission:
+  capacity_base_range: [1000, 4000]  # MW
+  capacity_uk_multiplier: 0.6  # UK connections typically lower
+  efficiency_range: [0.95, 0.99]
+  congestion_surcharge_range: [0.5, 5.0]  # EUR/MWh
+  fee_range: [0, 2.0]  # EUR/MWh

config/schema.yaml

@@ -169,18 +169,74 @@ schemas:
         type: "float32"
         unit: "TH/s"
         description: "Mining pool hashrate"
-      - name: "btc_price_usd"
+      - name: "btc_price_eur"
         type: "float32"
-        unit: "USD"
-        description: "Bitcoin price"
-      - name: "mining_profitability"
+        unit: "EUR"
+        description: "Bitcoin price in EUR"
+      - name: "power_efficiency_th_per_mw"
         type: "float32"
-        unit: "USD/TH/day"
-        description: "Mining profitability per terahash per day"
+        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 breakeven point"
+        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:
@@ -229,5 +285,32 @@ validation_rules:
   bitcoin_mining:
     - column: "hashrate_ths"
       min: 0
-    - column: "btc_price_usd"
+      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