Question

我正在尝试根据另一个列中是否包含字符串在pandas数据框中创建一个新列。我正在基于此post使用np.select。这是一个示例数据框和一个用于创建新列的示例函数

df=pd.DataFrame({'column':['one','ones','other','two','twos','others','three','threes']})

def add(df):

    conditions = [
        ('one' in df['column']),
        ('two' in df['column']),
        ('three' in df['column']),
        ('other' in df['column'])] 

    choices = [1, 2, 3, 0]
    df['Int'] = np.select(conditions, choices, default=0)

    return df

new_df=add(df)

我得到的输出是

   column  Int
0     one    0
1    ones    0
2   other    0
3     two    0
4    twos    0
5  others    0
6   three    0
7  threes    0

我想要的是

   column  Int
0     one    1
1    ones    1
2   other    0
3     two    2
4    twos    2
5  others    0
6   three    3
7  threes    3

我在做什么错了？

Answer 1

如果需要测试子字符串，请使用Series.str.contains：

 conditions = [
        (df['column'].str.contains('one')),
        (df['column'].str.contains('two')),
        (df['column'].str.contains('three')),
        (df['column'].str.contains('other'))]

如果需要完全匹配，请使用Series.eq或==：

 conditions = [
        (df['column'].eq('one')),
        (df['column'].eq('two')),
        (df['column'].eq('three')),
        (df['column'].eq('other'))]

 conditions = [
        (df['column'] == 'one'),
        (df['column'] == 'two'),
        (df['column'] == 'three'),
        (df['column'] == 'other')]

print (new_df)
   column  Int
0     one    1
1    ones    1
2   other    0
3     two    2
4    twos    2
5  others    0
6   three    3
7  threes    3

在np.select中使用条件字符串的问题

1 个答案: