实战场景：Pandas 如何统计某个数据列的空值个数
马上安排！
"""
对如下DF，设置两个单元格的值
·使用iloc 设置(3,B)的值是nan
·使用loc设置(8,D)的值是nan
"""
import numpy as np
import pandas as pd
np.random.seed(66)
df = pd.DataFrame(np.random.rand(10, 4), columns=list('ABCD'))
df.iloc[3, 1] = np.nan
df.loc[8, 'D'] = np.nan
print(df)
print(df.isnull().sum())A B C D
0 0.154288 0.133700 0.362685 0.679109
1 0.194450 0.251210 0.758416 0.557619
2 0.514803 0.467800 0.087176 0.829095
3 0.298641 NaN 0.678006 0.903489
4 0.514451 0.539105 0.664328 0.634057
5 0.353419 0.026643 0.165290 0.879319
6 0.067820 0.369086 0.115501 0.096294
7 0.083770 0.086927 0.022256 0.771043
8 0.049213 0.465223 0.941233 NaN
9 0.361318 0.031319 0.304045 0.188268
A 0
B 1
C 0
D 1
dtype: int64
1.创建有空值的DataFrame
import numpy as np
import pandas as pd
dates = pd.date_range("20200307", periods=4)
df1 = pd.DataFrame(np.arange(12).reshape(4, 3), index=dates, columns=["A", "B", "C"])
df2 = pd.DataFrame(df1, index=dates, columns=["A", "B", "C", "D"]) # 新增D列，却不赋值，NaN表示空值
print(df2)
# 打印输出：
# A B C D
# 2020-03-07 0 1 2 NaN
# 2020-03-08 3 4 5 NaN
# 2020-03-09 6 7 8 NaN
# 2020-03-10 9 10 11 NaN2.检查是否有空值
print(df2.isnull()) # 是空值返回True，否则返回False print(np.any(df2.isnull())) # 只要有一个空值便会返回True，否则返回False print(np.all(df2.isnull())) # 全部值都是空值便会返回True，否则返回False # 输出结果： # A B C D # 2020-03-07 False False False True # 2020-03-08 False False False True # 2020-03-09 False False False True # 2020-03-10 False False False True # True # False
3.给NaN赋值
df2.iloc[0, 3] = 10 # 直接给某个位置赋值 print(df2) # 打印输出： # A B C D # 2020-03-07 0 1 2 10.0 # 2020-03-08 3 4 5 NaN # 2020-03-09 6 7 8 NaN # 2020-03-10 9 10 11 NaN series = pd.Series([11, 12, 13], index=dates[1:4]) df2["D"] = series # 同时给D列赋多个值 print(df2) # 打印输出： # A B C D # 2020-03-07 0 1 2 NaN # 2020-03-08 3 4 5 11.0 # 2020-03-09 6 7 8 12.0 # 2020-03-10 9 10 11 13.0
4.去除有空值的行或列
df2.loc["2020-03-10", ["A", "B", "C"]] = [11, 12, 15]
df2.fillna("null") # 把空值填充成null
# dropna(axis,how,subset)方法会删除有空值的行或列，
# axis为0是行，axis为1是列,
# how为any时该行或列只要有一个空值就会删除，all是全都是空值才删除
# subset是一个列表，指定某些列
df2.dropna(axis=0, how="any", subset=["A", "D"])
