如何解决 ValueError：Series 的真值不明确。使用 a.empty、a.bool()、a.item()、a.any() 或 a.all()

Question

这是我现在正在处理的数据集列的示例：

print (data)
     Credit Days
0             30
1   Cash & Carry
2   Cash & Carry
3             20
4             20
5             30
6             15
7             10
8             15
9   Cash & Carry
10            10
11            10
12            21
13  Cash & Carry
14            20
15            20

所以该列同时包含字符串和整数值。我必须将这些值转换为整数评级，并且必须将它们保存到新创建的列中，例如 credit_days_rating。为此我写了一个代码：

data = pd.read_csv('test.csv', engine='python')

data['Credit Days'].astype(str)
if data['Credit Days']=='Cash & Carry':
    data['credit_days_rating'] = 4
else :
    data['Credit Days'].astype(int)
    if (data['Credit Days']>= 10) & (data['Credit Days']< 19):
        data['credit_days_rating'] = 3
    elif (data['Credit Days']>= 20) & (data['Credit Days']< 29):
        data['credit_days_rating'] = 2 
    else :
        data['credit_days_rating'] = 1 

为此，我收到以下错误日志：

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-65-f6ecf070a2d4> in <module>()
      2 
      3 data['Credit Days'].astype(str)
----> 4 if (data['Credit Days']=='Cash & Carry'):
      5     data['credit_days_rating'] = 5
      6 else :

~/anaconda3/envs/tensorflow/lib/python3.5/site-packages/pandas/core/generic.py in __nonzero__(self)
   1119         raise ValueError("The truth value of a {0} is ambiguous. "
   1120                          "Use a.empty, a.bool(), a.item(), a.any() or a.all()."
-> 1121                          .format(self.__class__.__name__))
   1122 
   1123     __bool__ = __nonzero__

ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

新列应如下所示：

Answer 1

您可以使用numpy.select 按条件列表设置值，比较数值使用to_numeric 和errors='coerce' 将非数字转换为NaNs：

m1 = data['Credit Days']=='Cash & Carry'

s = pd.to_numeric(data['Credit Days'], errors='coerce')
m2 = (s>= 10) & (s< 19)
m3 = (s>= 20) & (s< 29)
masks = [m1,m2,m3]
vals = [4,3,2]
data['credit_days_rating'] = np.select(masks, vals, default=1)
print (data)
     Credit Days  credit_days_rating
0             30                   1
1   Cash & Carry                   4
2   Cash & Carry                   4
3             20                   2
4             20                   2
5             30                   1
6             15                   3
7             10                   3
8             15                   3
9   Cash & Carry                   4
10            10                   3
11            10                   3
12            21                   2
13  Cash & Carry                   4
14            20                   2
15            20                   2

Answer 2

这是一种方法。使用强制将字符串设置为 NaN

s = pd.Series([21,'Cash & Carry',10,20])

df = pd.DataFrame(s,columns=['Credit Days'])
df["credit_days_rating"] = 'NaN'
df.loc[df['Credit Days'] == 'Cash & Carry', 'credit_days_rating'] = 5
df.loc[(pd.to_numeric(df['Credit Days'], errors='coerce') >= 10) & (pd.to_numeric(df['Credit Days'], errors='coerce') < 19),'credit_days_rating'] = 3

Answer 3

我猜你想要的实际上是在你的列上应用一个函数来获得一些只有整数类型的列。这可以通过以下方式完成：

data = ["some str", 10, 20, "some str", 1, 2, 3]
df = pd.DataFrame(data)

def my_function(value):
    if value == "some str":
        return 5
    elif value >= 10 or value < 19:
        return 3

df['new_col'] = df[0].apply(my_function)

df

那么输出是：

    0          new_col
0   some str    5
1   10          3
2   20          3
3   some str    5
4   1           3
5   2           3
6   3           3

(data['Credit Days']=='Cash & Carry') 正在做的是返回一个布尔值的 pandas Series 实例，例如：

df[0] == "some str"

0     True
1    False
2    False
3     True
4    False
5    False
6    False
Name: 0, dtype: bool

如果您想要一个布尔值，用于条件语句，您将需要使用 Series 内部方法 all() 或 any()：

(df[0] == "some str").any()

True