对熊猫数据框进行子集化时忽略 KeyError

Question

我有一个熊猫数据框 df，其中包含 city1、city2、city3、city4、city5 列。我有一个列表my_cities = ["city1","city3","city10"]。我想根据my_cities 中的列对 df 进行子集化。当我这样做时，

my_cities = ["city1","city3","city10"]

df_my_cities = df[my_cities]

我收到错误KeyError: "['city10'] not in index"

如果my_cities 中的元素不在df 中，我如何告诉代码继续执行？

Answer 1

您可以在所有列和list之间使用intersection：

df_my_cities = df[df.columns.intersection(my_cities)]

示例：

df = pd.DataFrame({'city1':['s', 'e'],
                   'city2':['e','f'],
                   'city3':['f','g'],
                   'city4':['r','g'],
                   'city5':['t','m']})

print (df)
  city1 city2 city3 city4 city5
0     s     e     f     r     t
1     e     f     g     g     m

my_cities = ["city1","city3","city10"]
df_my_cities = df[df.columns.intersection(my_cities)]
print (df_my_cities)
  city1 city3
0     s     f
1     e     g

或者numpy.intersect1d:

df_my_cities = df[np.intersect1d(df.columns, my_cities)]
print (df_my_cities)
  city1 city3
0     s     f
1     e     g