push

data/gen_nacc_meta.py

@@ -18,9 +18,33 @@ for k in dict_description:
     except:
         print(k)
 
+#%%
+df_tmp = pd.read_csv("./nacc_allowable_code.csv")
+df_tmp['variable_id'].fillna(method='ffill', inplace=True)
+
+dict_code = {}
+excluded_codes = set([])
+for k in dict_description.keys():
+    if k == 'his_LIVSIT':
+        nacc_name = 'NACCLIVS'
+    else:
+        nacc_name = k.split('_')[-1]
+    sub_df = df_tmp[df_tmp['variable_id'] == nacc_name]
+    dict_code[k] = {}
+    for i, row in sub_df.iterrows():
+        val = row['code_1']
+        if val in excluded_codes:
+            continue
+        description = row['descriptor']
+        dict_code[k][val] = description
+
+# df_code = pd.DataFrame.from_dict(dict_code, orient='index')
+
+
 # %%
 df_meta = pd.DataFrame()
 df_meta['Name'] = dict_description.keys()
 df_meta['Description'] = dict_description.values()
+df_meta['Values'] = dict_code.values()
 df_meta.to_csv('./input_meta_info.csv', index=False)
 # %%