pandas中的dropna()函数和np.dropna()函数

dropna()函数

参数：

axis: default 0指行,1为列,默认为0

how: {‘any’, ‘all’},‘any’指带缺失值的所有行;'all’指清除全是缺失值的,默认是any

thresh: int,保留含有int个非空值的行

subset: 对特定的列进行缺失值删除处理

inplace: 这个很常见,True表示直接在原数据上更改

简例

• data.dropna(how = ‘all’) # 丢弃全为缺失值的那些行
• data.dropna(axis = 1) # 丢弃有缺失值的列
• data.dropna(axis=1,how=“all”) # 丢弃全为缺失值的那些列
• data.dropna(axis=0,subset = [“Age”, “Sex”]) # 丢弃‘Age’和‘Sex’这两列中有缺失值的行

示例

构造个带缺失值的DataFrame

import pandas as pd
import numpy as np

data = pd.DataFrame(np.random.randn(5,4),index=list('abcde'),columns=['col1','col2','col3','col4'])

data

	col1	col2	col3	col4
a	0.757299	-0.641018	-1.471744	-1.200730
b	-0.902681	1.244215	0.399760	-1.536825
c	-1.175376	-1.168308	-0.086006	-0.237372
d	-0.552627	-0.096287	0.121881	0.457818
e	0.796096	1.720318	-1.758990	-1.870864

# 构造缺失值_1
data.iloc[1,:-1] = np.nan # iloc是根据索引进行操作

data

	col1	col2	col3	col4
a	0.757299	-0.641018	-1.471744	-1.200730
b	NaN	NaN	NaN	-1.536825
c	-1.175376	-1.168308	-0.086006	-0.237372
d	-0.552627	-0.096287	0.121881	0.457818
e	0.796096	1.720318	-1.758990	-1.870864

# 构造缺失值_2
data.iloc[1:-1,3] = np.nan

data

	col1	col2	col3	col4
a	0.757299	-0.641018	-1.471744	-1.200730
b	NaN	NaN	NaN	NaN
c	-1.175376	-1.168308	-0.086006	NaN
d	-0.552627	-0.096287	0.121881	NaN
e	0.796096	1.720318	-1.758990	-1.870864

对data进行测试

# 测试1
data.dropna() # 加上axis=0一样

	col1	col2	col3	col4
a	0.757299	-0.641018	-1.471744	-1.200730
e	0.796096	1.720318	-1.758990	-1.870864

# 测试2
data.dropna(axis=1) 

a
b
c
d
e

data.dropna(how='all')  # 只删除了全为nan的行

	col1	col2	col3	col4
a	0.757299	-0.641018	-1.471744	-1.200730
c	-1.175376	-1.168308	-0.086006	NaN
d	-0.552627	-0.096287	0.121881	NaN
e	0.796096	1.720318	-1.758990	-1.870864

data.dropna(thresh=1) # 只有1个nan的行被保留了

	col1	col2	col3	col4
a	0.757299	-0.641018	-1.471744	-1.200730
c	-1.175376	-1.168308	-0.086006	NaN
d	-0.552627	-0.096287	0.121881	NaN
e	0.796096	1.720318	-1.758990	-1.870864

data.dropna(axis=0,subset = ['col2','col3'])  # 删除了col2和col3这两列中有nan的行

	col1	col2	col3	col4
a	0.757299	-0.641018	-1.471744	-1.200730
c	-1.175376	-1.168308	-0.086006	NaN
d	-0.552627	-0.096287	0.121881	NaN
e	0.796096	1.720318	-1.758990	-1.870864