如何使用 Python 合并或使用字符串连接两个熊猫数据框

Question

我想合并两个数据框的数据。数据框是字符串

df1=pd.read_csv('test1.csv',encoding='utf8',index_col=['id_df1'],header=0)
df2=pd.read_csv('test2.csv',encoding='utf8',index_col=['id_df2'],header=0)
print(df1)
print(df2)

out:
id_df1  student  contact_person
1       john     Amy
2       jeff     Cindy
3       steven   Bob
4       tina     Amy


id_df2  student  parents_list
1       tina     (Amy) (Bob)
2       steven   (Eric) (Bob)
3       john     (Amy)
4       jeff     (Frank) (Harry)

print(type(df1['contact_person'][0]))
print(type(df2['parents_list'][0]))

out:
<class 'str'>
<class 'str'>

如果 df1['student']==df2['student'] & df1['contact_person'] 匹配 df2['parents_list']
我想使用这样的“外部”方法：

output

    id_df1 id_df2 student contact_person parents
0   1      3      john    Amy            (Amy)
1   3      2      steven  Bob            (Eric) (Bob)
2   4      1      tina    Amy            (Amy) (Bob)
3   2      null   jeff    Cindy          null
4   null   4      jeff    null           (Frank) (Harry)

Answer 1

一种方法是：首先在您的df1 中，如果满足您的条件，则创建一个列bool：

df1['bool'] = df1.apply(lambda row: True if row['contact_person'] in df2['parents_list'][df2['student'] == row['student']].iloc[0] else False,1)

那么你可以merge满足条件的df_yes和不满足条件的appenddf_no：

df_yes = df1[df1['bool'] == True].\
         merge(df2, on='student', how = 'left').drop('bool',1)
df_no = df1[df1['bool'] == False].\
        append(df2[df2['student'].isin(df1['student'][df1['bool'] == False])]).drop('bool',1)

最后只是append两者：

list_ordered_col = ['id_df1', 'id_df2', 'student', 'contact_person', 'parents_list']
df_output = df_yes.append(df_no)[list_ordered_col ].\
            reset_index(drop=True)

注意：它正在使用您之前对 parent_list 的输入（当它是 a、b、c...时）

编辑：将df1['bool']=... 替换为：

def parantes_in_parentList (row, df_list):
    df_parent_list = df_list['parents_list'][df_list['student'] == row['student']]
    if not df_parent_list.empty:
        if row['contact_person'] in df_parent_list.iloc[0]:
            return True
    # return False in all the other case
    return False
df1['bool'] = df1.apply(parantes_in_parentList , args=([df2]),axis=1)