Python-如果满足条件则应用公式

时间:2019-02-21 14:08:14

标签: python dataframe lambda

如果值以某些开头,我正在尝试更改数据框中的值。 我正在检查前4个值是否为0.00 如果以0.00开头,我想将该值乘以100 下面的公式给我这个错误

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

我的公式是

Total['Rate']=Total['Rate'].apply(lambda x: Total['Rate']*100 if \ 
Total['Rate'].str[:4]=='0.00' else Total['Rate'])

3 个答案:

答案 0 :(得分:3)

您是如此接近,问题在于在lambda函数中,您试图乘以整列而不是仅乘以值。

通过将它们更改为x(如下所示),就可以了。

Total['Rate'] = Total['Rate'].apply(lambda x: x*100 if str(x)[:4]=='0.00' else x, 1)

希望这会有所帮助!

答案 1 :(得分:1)

不必转换为字符串,最好将乘数值转换为integers并与0进行比较:

Total = pd.DataFrame(data=[0.001,0.2,5,0.0002],columns=['Rate'])

s = Total['Rate'] * 100
Total['Rate'] = np.where(s.astype(int) == 0, s, Total['Rate'])
print (Total)
   Rate
0  0.10
1  0.20
2  5.00
3  0.02

详细信息

print (s)
0      0.10
1     20.00
2    500.00
3      0.02
Name: Rate, dtype: float64

print (s.astype(int))
0      0
1     20
2    500
3      0
Name: Rate, dtype: int32

print (s.astype(int) == 0)
0     True
1    False
2    False
3     True
Name: Rate, dtype: bool

性能

Total = pd.DataFrame(data=[0.001,0.2,5,0.0002],columns=['Rate'])
Total = pd.concat([Total] * 10000, ignore_index=True)


In [296]: %%timeit
     ...: s = Total['Rate'] * 100
     ...: Total['Rate'] = np.where(s.round() == 0, s, Total['Rate'])
     ...: 
2.09 ms ± 119 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

In [297]: %%timeit
     ...: Total['Rate'] = Total['Rate'].apply(lambda x: x*100 if str(x)[:4]=='0.00' else x, 1)
     ...: 
26.2 ms ± 1.11 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)

编辑:如果要将值设置为多个掩码,例如否定0,请使用numpy.select

Total = pd.DataFrame(data=[0.001,0.2,5,0.0002, -10],columns=['Rate'])

s = Total['Rate'] * 100

mask1 = s.astype(int) == 0
mask2 = Total['Rate'] < 0

Total['Rate'] = np.select([mask1, mask2], [s, 0], default=Total['Rate'])
print (Total)
   Rate
0  0.10
1  0.20
2  5.00
3  0.02
4  0.00

答案 2 :(得分:0)

改为使用此:

Total['Rate']=Total['Rate'].mask(Total['Rate'].str.startswith('0.00'), Total['Rate']*100)