为什么将函数应用于数据框时会出现错误？

Question

def function(s):
    if (s['col1'] == 'something1')|(s['col1'] == 'smth2')|(s['col1'] == 'smth3'):
        return 'A'
    elif (s['col1'] == 'smth4')|(s['col1'] == 'smth5'):
        return 'B'
    elif (s['col1'] == 'smth6')|(s['col1'] == 'smth7'):
        return 'C'
    else:
        return 'D'

上面的功能有效。但是当我将它应用到数据框时：

df['new_col'] = df.apply(function, axis = 1)

我明白了：

TypeError: ("'bool' object is not callable", 'occurred at index 0') 

Answer 1

对我来说工作正常，这里是 Series.isin 和 numpy.select 的替代解决方案：

df = pd.DataFrame({
    'col1':['something1','jeff bridges','smth7','billy boy','smth5']})


print (df)

def function(s):
    if (s['col1'] == 'something1')|(s['col1'] == 'smth2')|(s['col1'] == 'smth3'):
        return 'A'
    elif (s['col1'] == 'smth4')|(s['col1'] == 'smth5'):
        return 'B'
    elif (s['col1'] == 'smth6')|(s['col1'] == 'smth7'):
        return 'C'
    else:
        return 'D'
    
df['new_col'] = df.apply(function, axis = 1)

---

m1 = df['col1'].isin(['something1','smth2','smth3'])
m2 = df['col1'].isin(['smth4','smth5'])
m3 = df['col1'].isin(['smth6','smth7'])

df['new_col1'] = np.select([m1, m2, m3], ['A','B','C'], default='D')
print (df)
           col1 new_col new_col1
0    something1       A        A
1  jeff bridges       D        D
2         smth7       C        C
3     billy boy       D        D
4         smth5       B        B