Feature: Expanded column mapping with NLP variants and extra AML fields

modules/etl.py

@@ -3,30 +3,41 @@ import numpy as np
 
 # Column mapping — maps ANY known variant to internal name
 COLUMN_MAP = {
-    # transaction_id variants
+    # Normalize column names first (lowercase + strip)
+    "transaction id": "transaction_id",
     "tx_id": "transaction_id",
     "txn_id": "transaction_id",
 
-    # customer_id — ONLY map sender, not receiver
-    "sender_account_id": "customer_id",   # ← sender is the customer
-    "account_id": "customer_id",
+    # customer
+    "person involved": "customer_id",
+    "sender_account_id": "customer_id",
     "cust_id": "customer_id",
 
-    # Keep receiver separate — DO NOT map to customer_id
-    # "receiver_account_id" stays as receiver_account_id
-
-    # amount variants
+    # amount — handles spaces and special chars
+    "amount (usd)": "amount",
     "tx_amount": "amount",
     "txn_amount": "amount",
 
-    # timestamp variants
+    # timestamp
+    "date of transaction": "timestamp",
     "transaction_date": "timestamp",
     "tx_date": "timestamp",
 
-    # transaction_type variants
+    # transaction type
+    "transaction type": "transaction_type",
     "tx_type": "transaction_type",
     "txn_type": "transaction_type",
-    "type": "transaction_type",
+
+    # countries
+    "country": "origin_country",
+    "destination country": "dest_country",
+
+    # extra AML fields
+    "money laundering risk score": "ml_risk_score",
+    "shell companies involved": "shell_companies",
+    "tax haven country": "tax_haven",
+    "source of money": "source_of_money",
+    "reported by authority": "reported_by_authority",
 }
 
 def load_and_validate(file):