Add function to create ISCO-08 hierarchy from CSV file

isco.py

@@ -0,0 +1,39 @@
+# filename: isco.py
+import csv
+
+
+def create_hierarchy(filename):
+    """
+    Creates a dictionary where keys are nodes and values are sets of parent nodes representing the hierarchy of the ISCO-08 codes from the "unit" column of the isco_structure.csv file.
+
+    Args:
+    - filename: A string representing the path to the CSV file containing the ISCO-08 codes and their hierarchy.
+
+    Returns:
+    - A dictionary where keys are ISCO-08 unit codes and values are sets of their parent codes.
+    """
+    isco_hierarchy = {}
+
+    with open(filename, newline="") as csvfile:
+        reader = csv.DictReader(csvfile)
+        for row in reader:
+            # Extract unit group level code (4 digits)
+            unit_code = row["unit"].zfill(4)
+            # Extract the parent code for the unit group level, which is the minor group level (3 digits)
+            parent_code = unit_code[:3]
+
+            # Add the unit code to the hierarchy with its parent code
+            isco_hierarchy[unit_code] = {parent_code}
+
+            # Additionally, we can add the parent's parent codes if needed
+            # For example, the major group level (1 digit) and sub-major group level (2 digits)
+            major_code = unit_code[0]
+            sub_major_code = unit_code[:2]
+            isco_hierarchy[unit_code].update({major_code, sub_major_code})
+
+    return isco_hierarchy
+
+
+# Example usage:
+# hierarchy = create_hierarchy("ISCO_structure.csv")
+# print(hierarchy)