python - np.where 多个变量

Question

我有一个数据框:

customer_id [1,2,3,4,5,6,7,8,9,10]
feature1 [0,0,1,1,0,0,1,1,0,0]
feature2 [1,0,1,0,1,0,1,0,1,0]
feature3 [0,0,1,0,0,0,1,0,0,0]

使用这个我想创建一个新变量(比如new_var)来表示当特征1为1时new_var=1，如果feature_2=1则new_var=2，feature3=1则new_var=3，否则4。我是尝试 np.where 但虽然它不会给我错误，但它没有做正确的事情 - 所以我猜嵌套的 np.where 仅适用于单个变量。在这种情况下，在 pandas 中执行嵌套 if/case 的最佳方法是什么？

我的 np.where 代码是这样的:

df[new_var]=np.where(df['feature1']==1,'1', np.where(df['feature2']==1,'2', np.where(df[feature3']==1,'3','4')))

Answer 1

我认为你需要numpy.select - 它选择第一个 True 值，其他所有值都不重要:

m1 = df['feature1']==1 
m2 = df['feature2']==1    
m3 = df['feature3']==1 
df['new_var'] = np.select([m1, m2, m3], ['1', '2', '3'], default='4')

示例:

customer_id = [1,2,3,4,5,6,7,8,9,10]
feature1 = [0,0,1,1,0,0,1,1,0,0]
feature2 = [1,0,1,0,1,0,1,0,1,0]
feature3  = [0,0,1,0,0,0,1,0,0,0]

df = pd.DataFrame({'customer_id':customer_id,
                   'feature1':feature1,
                   'feature2':feature2,
                   'feature3':feature3})

m1 = df['feature1']==1 
m2 = df['feature2']==1    
m3 = df['feature3']==1 
df['new_var'] = np.select([m1, m2, m3], ['1', '2', '3'], default='4')
print (df)
   customer_id  feature1  feature2  feature3 new_var
0            1         0         1         0       2
1            2         0         0         0       4
2            3         1         1         1       1
3            4         1         0         0       1
4            5         0         1         0       2
5            6         0         0         0       4
6            7         1         1         1       1
7            8         1         0         0       1
8            9         0         1         0       2
9           10         0         0         0       4

如果在features中只有1和0可以将0转换为False 和 1 为 True:

m1 = df['feature1'].astype(bool)
m2 = df['feature2'].astype(bool)
m3 = df['feature3'].astype(bool)
df['new_var'] = np.select([m1, m2, m3], ['1', '2', '3'], default='4')
print (df)
   customer_id  feature1  feature2  feature3 new_var
0            1         0         1         0       2
1            2         0         0         0       4
2            3         1         1         1       1
3            4         1         0         0       1
4            5         0         1         0       2
5            6         0         0         0       4
6            7         1         1         1       1
7            8         1         0         0       1
8            9         0         1         0       2
9           10         0         0         0       4