缺失值处理

缺失值处理
import numpy as np
import pandas as pd
data = np.array([1,2,5,4,np.nan])
print(data)
print(‘numpy 有专门的nansum，nanmax等操作处理缺失值‘)
# r = data.sum()
r = np.nansum(data)
print(r)

print(‘pandas 将None和np.nan都处理为NaN‘)
data = pd.Series([2,3,None, 4,5,np.nan])
print(data)

print(‘-‘*10,‘获取缺失值‘,‘-‘*10)
index = data.isnull()
print(data[index])

print(‘-‘*10,‘剔除缺失值‘,‘-‘*10)
df = pd.DataFrame([[1,2,np.nan],
                   [4,None,5],
                   [6,7,8]])
print(df)

dd = df.dropna()
print(dd)
print(‘原始的df不会去掉nan，需要接一下‘)
print(‘剔除缺失值的列‘)
dd = df.dropna(axis = 1)#等价于axis=‘columns’
print(dd)

print(‘how="any",只要存在缺失值就剔除；how=all，全部都是缺失值才会剔除‘)
dd = df.dropna(how="any")
print(dd)

#缺失值填充
import numpy as np
import pandas as pd
df = pd.DataFrame([[1,2,np.nan],
                   [4,None,5],
                   [6,7,8]], index=list(‘abc‘),columns=list(‘ABC‘))
print(‘-‘*10,‘原始df数据‘,‘-‘*10)
print(df)
print(‘-‘*10,‘缺失值填充默认值0‘,‘-‘*10)
dd = df.fillna(0)
print(dd)
print(‘-‘*10,‘缺失值填充前一行元素‘,‘-‘*10)
dd = df.fillna(method=‘ffill‘)
print(dd)
print(‘-‘*10,‘缺失值填充后一行元素‘,‘-‘*10)
dd = df.fillna(method=‘bfill‘)
print(dd)
print(‘-‘*10,‘缺失值填充前一列元素‘,‘-‘*10)
dd = df.fillna(method=‘ffill‘,axis = 1)
print(dd)
print(‘-‘*10,‘缺失值填充后一列元素‘,‘-‘*10)
dd = df.fillna(method=‘bfill‘,axis = 1)
print(dd)
print(‘-‘*10,‘按列的平均值进行填充‘,‘-‘*10)
for i in df.columns:#df.index就是按照行求平均值
    fill = np.nanmean(df[i])
    dd[i]=df[i].fillna(fill)
print(dd)