Income decile fethcing

data_processing/age_distribution_by_id.py

@@ -34,6 +34,7 @@ def get_age_distribution(years, smsv_ids, filename):
 
                 if row_smsv_id not in age_distributions:
                     age_distributions[row_smsv_id] = {}
+
                 if row_year not in age_distributions[row_smsv_id]:
                     age_distributions[row_smsv_id][row_year] = {}
                 
@@ -47,7 +48,8 @@ if __name__ == '__main__':
     # Example usage:
     filename = os.path.join('given_data', 'ibuafjoldi.csv')
     smsv_ids_to_find = ['0103', '2903', '4002']  # List of desired smsv_ids
-    age_dist = get_age_distribution([2017, 2024],smsv_ids_to_find, filename)
+    years = [2017, 2024]
+    age_dist = get_age_distribution(years, smsv_ids_to_find, filename)
 
     for smsv_id, distributions in age_dist.items():
         print(f"Age Distribution for SMSV ID: {smsv_id}")

data_processing/income_decile_by_id.py

@@ -1,4 +1,5 @@
 import csv
+import os
 
 def open_file(filename):
     try:
@@ -14,17 +15,42 @@ def open_file(filename):
         return []
     
 def get_income_decile(years, smsv_ids, filename):
-    income_deciles = {smsv_ids: {year: {} for year in years} for smsv_id in smsv_ids} # Initialize output dict
+    income_deciles = {smsv_id: {year: {} for year in years} for smsv_id in smsv_ids} # Initialize output dict
     csv_data = open_file(filename)
 
-    if csv_data(): # Check if data is empty
+    if csv_data: # Check if data is empty
         header = csv_data[0] # Extract header for column indexing
         smsv_id_index = header.index('smasvaedi')
         decile_index = header.index('tekjutiund')
-           
-    
+        population_index = header.index('fjoldi')
+        year_index = header.index('ar')
+
+        for row in csv_data[1:]: # Skip header
+            row_smsv_id = row[smsv_id_index]
+            row_year = int(row[year_index])
+            if row_smsv_id in smsv_ids and row_year in years:
+                decile_group = row[decile_index]
+                population = int(row[population_index])
+
+                if row_smsv_id not in income_deciles:
+                    income_deciles[row_smsv_id] = {}
+
+                if row_year not in income_deciles[row_smsv_id]:
+                    income_deciles[row_smsv_id][row_year] = {}
+
+                if decile_group not in income_deciles[row_smsv_id][row_year]:
+                    income_deciles[row_smsv_id][row_year][int(decile_group)] = population
+    return income_deciles
 
 if __name__ == '__main__':
     # Example usage:
-    filename = './given_data/ibuafjoldi.csv'
-    smsv_ids_to_find = ['0103', '2903', '4002']  # List of desired smsv_ids
+    filename = os.path.join('given_data', 'tekjutiundir.csv')
+    smsv_ids_to_find = ['0103', '2903', '4002']  # List of desired smsv_ids
+    years = [2017, 2024]
+    income_decile = get_income_decile(years, smsv_ids_to_find, filename)
+
+    for smsv_id, deciles in  income_decile.items():
+        print(f"Income Deciles for SMSV ID: {smsv_id}")
+        for decile, population in deciles.items():
+            print(f" - {decile}: {population}")
+        print()