SetDiagCodes.py

def create_map():
    ## List of tuples with name and number of repititons.
    name_list = [('infections', 139),
                ('neoplasms', (239 - 139)),
                ('endocrine', (279 - 239)),
                ('blood', (289 - 279)),
                ('mental', (319 - 289)),
                ('nervous', (359 - 319)),
                ('sense', (389 - 359)),
                ('circulatory', (459-389)),
                ('respiratory', (519-459)),
                ('digestive', (579 - 519)),
                ('genitourinary', (629 - 579)),
                ('pregnancy', (679 - 629)),
                ('skin', (709 - 679)),
                ('musculoskeletal', (739 - 709)),
                ('congenital', (759 - 739)),
                ('perinatal', (779 - 759)),
                ('ill-defined', (799 - 779)),
                ('injury', (999 - 799))]
    ## Loop over the tuples to create a dictionary to map codes 
    ## to the names.
    out_dict = {}
    count = 1
    for name, num in name_list:
        for i in range(num):
          out_dict.update({str(count): name})  
          count += 1
    return out_dict
  

def map_codes(df, codes):
    import pandas as pd
    col_names = df.columns.tolist()
    for col in col_names:
        temp = [] 
        for num in df[col]:           
            if ((num is None) | (num in ['unknown', '?']) | (pd.isnull(num))): temp.append('unknown')
            elif(num.upper()[0] == 'V'): temp.append('supplemental')
            elif(num.upper()[0] == 'E'): temp.append('injury')
            else: 
                lkup = num.split('.')[0]
                temp.append(codes[lkup])           
        df.loc[:, col] = temp               
    return df 

def azureml_main(df):
    col_list = ['diag_1', 'diag_2', 'diag_3']
    codes = create_map()
    df[col_list] = map_codes(df[col_list], codes)
    return df