通过从邻居中随机选择一个邮政编码来填充缺失的邮政编码

Question

我有一个像下面这样的pandas数据框，我正在尝试通过从类似的neighbourhood_group_cleansed中选择任何随机值来替换ZipCode域中缺少的值。下面是我的尝试，但效果不是很好。请帮帮忙。

zipcodes = a_df[['neighbourhood_group_cleansed','zipcode']].drop_duplicates().reset_index()
a_df['zipcode'] = a_df.apply(lambda row: np.random.choice(zipcodes[zipcodes['neighbourhood_group_cleansed'] == 
                row['neighbourhood_group_cleansed']]['zipcode']) if len(row.zipcode) == 0   else row.zipcode, axis = 1)

state   city    smart_location  neighbourhood_group_cleansed    zipcode
0   NY  New York    New York, NY    Manhattan   10029
1   NY  Brooklyn    Brooklyn, NY    Brooklyn    11221
2   NY  Brooklyn    Brooklyn, NY    Brooklyn    11206
3   NY  New York    New York, NY    Manhattan   10001
4   NY  New York    New York, NY    Manhattan   10162
... ... ... ... ... ...
6492    NY  New York    New York, NY    Manhattan   10004.0
6493    NY  Brooklyn    Brooklyn, NY    Brooklyn    11229.0
6494    NY  Queens  Queens, NY  Queens  11691.0
6495    NY  New York    New York, NY    Manhattan   10044.0
6496    NY  Brooklyn    Brooklyn, NY    Brooklyn    11234.0

Answer 1

这应该是可行的

df['zipcode'] = df.apply(lambda x: random.choice(df[df['neighbourhood_group_cleansed'] == x['neighbourhood_group_cleansed']].zipcode.dropna().values) if np.isnan(x['zipcode']) else x['zipcode'], axis=1)