python - 从数据框python创建自定义字典时出现字符串索引错误

Question

我正在尝试从数据框创建字典，下面是数据框和代码:

Code | Desc
XS   | Train
XS   | Car
SE   | Cycle
SE   | Train

下面是我的代码

lst_code = 'NA'
comp_list=[]
comp_dict = {}
for row in test_df:
    if str(row['code']) != lst_code:
        lst_code = row['code']
        if comp_list:
            comp_dict.update(lst_code,comp_list)
    else:
        comp_list.append(row['desc'])

使用上面的代码我得到以下错误

if str(row['analyst_code']) != lst_code:
TypeError: string indices must be integers

我期待下面的字典:

comp_dict = {'XS':['Train','Car'],
          'SE':['Cycle','Train']}

请建议我该如何解决这个问题？

Answer 1

首先按 boolean indexing 过滤然后按 GroupBy.size 每组计数, 最后转换 Series to_dict :

lst_code = 'NA'
comp_dict = df[df['Code'] != lst_code].groupby('Code')['Desc'].apply(list).to_dict()
print (comp_dict)
{'SE': ['Cycle', 'Train'], 'XS': ['Train', 'Car']}

如果不需要过滤:

comp_dict = df.groupby('code')['Desc'].apply(list).to_dict()