numpy笔记：.select()默认值的类型

numpy.select(condlist, choicelist, default=0)链接，

当DataFame新生成字段时，choicelist如下3种会使得default的“0”的type分别为str和int：

一、当select函数里choicelist是指定string value时，type为str：

f['new'] = numpy.select([mask], ['value'], default=0)

二、当select函数里choicelist是df.series时，type为数值：

f['new'] = numpy.select([mask], [df.Series], default=0)

三、当select函数里choicelist是int / float时，type为数值：

f['new'] = numpy.select([mask], [int], default=0)

import pandas as pd
import numpy as np

df= pd.DataFrame({'col1': ['b', 'str', 1, 1],
                 'col2': ['str', 12, 5, 41]})
# pinrt(df)
	col1	col2
0	b	22
1	str	12
2	1	5
3	1	41

m1 = d.col1 == 'b'
df['t1'] = np.select([m1], ['22'], default=0)
df['t2'] = np.select([m1], [df['col2']], default=0)
df['t3'] = np.select([m1], [0.1], default=0)
for i in ['t1', 't2', 't3']:
    print(any(df[f'{i}']==0))
    print(any(df[f'{i}']=='0'))
# 结果如下：
False
True

True
False

True
False

当default=np.nan时候：

m1 = d.col1 == 'b'
df['t1'] = np.select([m1], ['22'], default=np.nan)
df['t2'] = np.select([m1], [df['col2']], default=np.nan)
df['t3'] = np.select([m1], [0.1], default=np.nan)
for i in ['t1', 't2', 't3']:
    print(any(df[f'{i}']=='nan'))
    print(any(df[f'{i}'].isna()))
    print(any(df[f'{i}'].isna()))

# 结果如下
True
False
False

False
True
True

False
True
True