Upload location_standardization.py

location_standardization.py

@@ -0,0 +1,20 @@
+# location_standardization.py
+
+from rapidfuzz import process, fuzz
+
+def group_similar_locations(locations, threshold=85):
+    clustered = {}
+    canonical_map = {}
+
+    for loc in locations:
+        if not loc or not isinstance(loc, str): continue
+        loc = loc.strip()
+        match, score, _ = process.extractOne(loc, clustered.keys(), scorer=fuzz.ratio)
+        if score >= threshold:
+            canonical_map[loc] = match
+            clustered[match].append(loc)
+        else:
+            clustered[loc] = [loc]
+            canonical_map[loc] = loc
+
+    return canonical_map