假设我有一个交易和客户数据框:
df = pd.DataFrame({'shop': pd.Series(['McDonalds', 'McDonalds', 'McDonalds', 'McDonalds', 'Burger King', 'Burger King', 'Burger King', 'Burger King', 'Burger King', 'Trump Golf Course', 'Trump Golf Course', 'Trump Golf Course', 'Trump Golf Course', 'Trump Golf Course', 'Trump Golf Course'],dtype='object',index=pd.RangeIndex(start=0, stop=15, step=1)), 'Customer': pd.Series(['John Ryan', 'Jim Bob', 'Mary Ryan', 'Michael Patric', 'John Ryan', 'Jim Bob', 'Mary Ryan', 'Sean Connery', 'Brad Pitt', 'John Ryan', 'John Ryan', 'Michael Patric', 'Mary Ryan', 'John Ryan', 'Jim Bob'],dtype='object',index=pd.RangeIndex(start=0, stop=15, step=1)), 'Customer ID': pd.Series([1, 2, 3, 4, 1, 2, 3, 5, 6, 1, 1, 4, 3, 1, 2],dtype='int64',index=pd.RangeIndex(start=0, stop=15, step=1)), 'Amount': pd.Series([50, 32, 15, 65, 32, 51, 54, 84, 52, 51, 2, 32, 54, 87, 65],dtype='int64',index=pd.RangeIndex(start=0, stop=15, step=1))}, index=pd.RangeIndex(start=0, stop=15, step=1))
print(df)
shop Customer Customer ID Amount
0 McDonalds John Ryan 1 50
1 McDonalds Jim Bob 2 32
2 McDonalds Mary Ryan 3 15
3 McDonalds Michael Patric 4 65
4 Burger King John Ryan 1 32
5 Burger King Jim Bob 2 51
6 Burger King Mary Ryan 3 54
7 Burger King Sean Connery 5 84
8 Burger King Brad Pitt 6 52
9 Trump Golf Course John Ryan 1 51
10 Trump Golf Course John Ryan 1 2
11 Trump Golf Course Michael Patric 4 32
12 Trump Golf Course Mary Ryan 3 54
13 Trump Golf Course John Ryan 1 87
14 Trump Golf Course Jim Bob 2 65
我想提取或标记未在麦当劳购物的汉堡王顾客。 (在这种情况下，肖恩康纳利和布拉德皮特)
我正在尝试在 shop == McDonalds 处创建一个掩码，并获取客户 ID
mask1 = df.shop == 'McDonalds'
mcdonalds_customer_ids = df[mask1]['Customer ID'].values
array([1, 2, 3, 4], dtype=int64)
然后创建一个单独的掩码，其中 shop=='Burger King' 并且客户 ID 不在麦当劳客户 ID 列表中:
mask = (df['shop'] == 'Burger King' & df['Customer ID'] not in mcdonalds_customer_ids)
我收到以下错误:
TypeError: ufunc 'bitwise_and' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''
TypeError: cannot compare a dtyped [int64] array with a scalar of type [bool]
我也尝试过使用 np.where，但它变得更加困惑。
我的预期输出只是提取两个没有在麦当劳购物的汉堡王顾客:
shop Customer Customer ID Amount
7 Burger King Sean Connery 5 84
8 Burger King Brad Pitt 6 52
或者使用 np.where 标记它们:
shop Customer Customer ID Amount No_McDonalds
7 Burger King Sean Connery 5 84 True
8 Burger King Brad Pitt 6 52 True
我可以用一个函数来做到这一点，但希望以某种方式对其进行矢量化。完全失败，任何帮助表示赞赏。
请您参考如下方法:
这里有一个向量化的解决方案 loc和 isin . 首先我们获取 Burger King 的行，然后我们从 McDonalds
Customers
最后检查汉堡王的哪些顾客没有去过麦当劳:
bk = df.loc[df['shop'].eq('Burger King')]
mc = df.loc[df['shop'].eq('McDonalds'), 'Customer']
bk[~bk['Customer'].isin(mc)]
shop Customer Customer ID Amount
7 Burger King Sean Connery 5 84
8 Burger King Brad Pitt 6 52
