• 28 comments
• Andrew Coyne

10. 缺失值处理

import numpy as np
import pandas as pd
rng = np.random.default_rng(0)

df = pd.DataFrame(
#     rng.random((5,3)),
    np.random.randn(5, 3),
    ['a','c','e','f','h'],
    ['one','two','three']
)
df

df['four'] = 'bar'
df

df['five'] = df['one'] > 0
df

# 重置index
df2 = df.reindex(list('abcdefgh'))
df2

# 那些是空值
pd.isna(df2)

# 判断某一列(Series),
pd.isna(df2['one'])

# 那些不是空值
df2.notna()

df2['one'].notna()

# 人为制造一个空值,忽略警告.
df2['three']['e'] = np.nan

# 缺失值
df2.fillna(0)

# 中位数median(),每一个column计算一个中位数
df2.fillna(df2.median())

# 均值
df2.fillna(df2.mean())

# 删除
# 默认值 ,0轴只要有一个值为空,就删除
df2.dropna(axis=0,how='any')

# 删除, all 全部为空 any 任意一个为空
df2.dropna(how='all')

# 指定轴 0轴上有空值的,1轴上有空值的
df2.dropna(axis=0)
# df3.dropna(axis=0,how='any')

df3 = df2.dropna(how='all')
df3
# 删除的是column
df3.dropna(axis=1,how='any')

df3 = df2.dropna(how='all')
df3
df3.dropna(axis=1,how='all')