Python3.5 Pandas模块缺失值处理和层次索引实例详解

来源：jb51　　时间：2019/4/24 9:43:19　　对本文有异议

本文实例讲述了Python3.5 Pandas模块缺失值处理和层次索引。分享给大家供大家参考，具体如下：

1、pandas缺失值处理

import numpy as np
import pandas as pd
from pandas import Series,DataFrame
 
df3 = DataFrame([
  ["Tom",np.nan,456.67,"M"],
  ["Merry",34,345.56,np.nan],
  [np.nan,np.nan,np.nan,np.nan],
  ["John",23,np.nan,"M"],
  ["Joe",18,385.12,"F"]
],columns = ["name","age","salary","gender"])
 
print(df3)
print("=======判断NaN值=======")
print(df3.isnull())
print("=======判断非NaN值=======")
print(df3.notnull())
print("=======删除包含NaN值的行=======")
print(df3.dropna())
print("=======删除全部为NaN值的行=======")
print(df3.dropna(how="all"))
 
df3.ix[2,0] = "Gerry"    #修改第2行第0列的值
print(df3)
 
print("=======删除包含NaN值的列=======")
print(df3.dropna(axis=1))

运行结果：

   name   age salary gender
0    Tom   NaN 456.67      M
1 Merry 34.0 345.56    NaN
2    NaN   NaN     NaN    NaN
3   John 23.0     NaN      M
4    Joe 18.0 385.12      F
=======判断NaN值=======
    name    age salary gender
0 False   True False False
1 False False False   True
2   True   True   True   True
3 False False   True False
4 False False False False
=======判断非NaN值=======
    name    age salary gender
0   True False   True   True
1   True   True   True False
2 False False False False
3   True   True False   True
4   True   True   True   True
=======删除包含NaN值的行=======
name   age salary gender
4 Joe 18.0 385.12      F
=======删除全部为NaN值的行=======
    name   age salary gender
0    Tom   NaN 456.67      M
1 Merry 34.0 345.56    NaN
3   John 23.0     NaN      M
4    Joe 18.0 385.12      F
    name   age salary gender
0    Tom   NaN 456.67      M
1 Merry 34.0 345.56    NaN
2 Gerry   NaN     NaN    NaN
3   John 23.0     NaN      M
4    Joe 18.0 385.12      F
=======删除包含NaN值的列=======
    name
0    Tom
1 Merry
2 Gerry
3   John
4    Joe

import numpy as np
import pandas as pd
from pandas import Series,DataFrame
 
df4 = DataFrame(np.random.randn(7,3))
print(df4)
 
df4.ix[:4,1] = np.nan    #第0至3行，第1列的数据
df4.ix[:2,2] = np.nan
print(df4)
 
print(df4.fillna(0))    #将缺失值用传入的指定值0替换
 
print(df4.fillna({1:0.5,2:-1}))   #将缺失值按照指定形式填充

运行结果：

          0         1         2
0 -0.737618 -0.530302 -2.716457
1 0.810339 0.063028 -0.341343
2 0.070564 0.347308 -0.121137
3 -0.501875 -1.573071 -0.816077
4 -2.159196 -0.659185 -0.885185
5 0.175086 -0.954109 -0.758657
6 0.395744 -0.875943 0.950323
          0         1         2
0 -0.737618       NaN       NaN
1 0.810339       NaN       NaN
2 0.070564       NaN       NaN
3 -0.501875       NaN -0.816077
4 -2.159196       NaN -0.885185
5 0.175086 -0.954109 -0.758657
6 0.395744 -0.875943 0.950323
          0         1         2
0 -0.737618 0.000000 0.000000
1 0.810339 0.000000 0.000000
2 0.070564 0.000000 0.000000
3 -0.501875 0.000000 -0.816077
4 -2.159196 0.000000 -0.885185
5 0.175086 -0.954109 -0.758657
6 0.395744 -0.875943 0.950323
          0         1         2
0 -0.737618 0.500000 -1.000000
1 0.810339 0.500000 -1.000000
2 0.070564 0.500000 -1.000000
3 -0.501875 0.500000 -0.816077
4 -2.159196 0.500000 -0.885185
5 0.175086 -0.954109 -0.758657
6 0.395744 -0.875943 0.950323

2、pandas常用数学统计方法

import numpy as np
import pandas as pd
from pandas import Series,DataFrame
 
#pandas常用数学统计方法
 
arr = np.array([
  [98.5,89.5,88.5],
  [98.5,85.5,88],
  [70,85,60],
  [80,85,82]
])
df1 = DataFrame(arr,columns=["语文","数学","英语"])
print(df1)
print("=======针对列计算总统计值=======")
print(df1.describe())
print("=======默认计算各列非NaN值个数=======")
print(df1.count())
print("=======计算各行非NaN值个数=======")
print(df1.count(axis=1))

运行结果：

     语文    数学    英语
0 98.5 89.5 88.5
1 98.5 85.5 88.0
2 70.0 85.0 60.0
3 80.0 85.0 82.0
=======针对列计算总统计值=======
              语文         数学         英语
count   4.000000   4.000000   4.000000
mean   86.750000 86.250000 79.625000
std    14.168627   2.179449 13.412525
min    70.000000 85.000000 60.000000
25%    77.500000 85.000000 76.500000
50%    89.250000 85.250000 85.000000
75%    98.500000 86.500000 88.125000
max    98.500000 89.500000 88.500000
=======默认计算各列非NaN值个数=======
语文    4
数学    4
英语    4
dtype: int64
=======计算各行非NaN值个数=======
0    3
1    3
2    3
3    3
dtype: int64

import numpy as np
import pandas as pd
from pandas import Series,DataFrame、
 
#2.pandas相关系数与协方差
df2 = DataFrame({
  "GDP":[12,23,34,45,56],
  "air_temperature":[23,25,26,27,30],
  "year":["2001","2002","2003","2004","2005"]
})
 
print(df2)
print("=========相关系数========")
print(df2.corr())
print("=========协方差========")
print(df2.cov())
print("=========两个量之间的相关系数========")
print(df2["GDP"].corr(df2["air_temperature"]))
print("=========两个量之间协方差========")
print(df2["GDP"].cov(df2["air_temperature"]))

运行结果：

GDP air_temperature year
0   12               23 2001
1   23               25 2002
2   34               26 2003
3   45               27 2004
4   56               30 2005
=========相关系数========
                      GDP air_temperature
GDP              1.000000         0.977356
air_temperature 0.977356         1.000000
=========协方差========
                   GDP air_temperature
GDP              302.5             44.0
air_temperature   44.0              6.7
=========两个量之间的相关系数========
0.97735555485
=========两个量之间协方差========
44.0

import numpy as np
import pandas as pd
from pandas import Series,DataFrame
 
#3.pandas唯一值、值计数及成员资格
 
df3 = DataFrame({
  "order_id":["1001","1002","1003","1004","1005"],
  "member_id":["m01","m01","m02","m01","m02",],
  "order_amt":[345,312.2,123,250.2,235]
})
 
print(df3)
 
print("=========去重后的数组=========")
print(df3["member_id"].unique())
 
print("=========值出现的频率=========")
print(df3["member_id"].value_counts())
 
print("=========成员资格=========")
df3 = df3["member_id"]
mask = df3.isin(["m01"])
print(mask)
print(df3[mask])

运行结果：

member_id order_amt order_id
0       m01      345.0     1001
1       m01      312.2     1002
2       m02      123.0     1003
3       m01      250.2     1004
4       m02      235.0     1005
=========去重后的数组=========
['m01' 'm02']
=========值出现的频率=========
m01    3
m02    2
Name: member_id, dtype: int64
=========成员资格=========
0     True
1     True
2    False
3     True
4    False
Name: member_id, dtype: bool
0    m01
1    m01
3    m01
Name: member_id, dtype: object

3、pandas层次索引

import numpy as np
import pandas as pd
from pandas import Series,DataFrame
 
#3.pandas层次索引
data = Series([998.4,6455,5432,9765,5432],
       index=[["2001","2001","2001","2002","2002"],
       ["苹果","香蕉","西瓜","苹果","西瓜"]]
       )
print(data)
 
df4 = DataFrame({
  "year":[2001,2001,2002,2002,2003],
  "fruit":["apple","banana","apple","banana","apple"],
  "production":[2345,5632,3245,6432,4532],
  "profits":[245.6,432.7,534.1,354,467.8]
})
 
print(df4)
print("=======层次化索引=======")
df4 = df4.set_index(["year","fruit"])
print(df4)
print("=======依照索引取值=======")
print(df4.ix[2002,"apple"])
print("=======依照层次化索引统计数据=======")
print(df4.sum(level="year"))
print(df4.mean(level="fruit"))
print(df4.min(level=["year","fruit"]))

运行结果：

2001 苹果     998.4
      香蕉    6455.0
      西瓜    5432.0
2002 苹果    9765.0
      西瓜    5432.0
dtype: float64
    fruit production profits year
0   apple        2345    245.6 2001
1 banana        5632    432.7 2001
2   apple        3245    534.1 2002
3 banana        6432    354.0 2002
4   apple        4532    467.8 2003
=======层次化索引=======
             production profits
year fruit
2001 apple         2345    245.6
     banana        5632    432.7
2002 apple         3245    534.1
     banana        6432    354.0
2003 apple         4532    467.8
=======依照索引取值=======
production    3245.0
profits        534.1
Name: (2002, apple), dtype: float64
=======依照层次化索引统计数据=======
      production profits
year
2001        7977    678.3
2002        9677    888.1
2003        4532    467.8
        production     profits
fruit
apple         3374 415.833333
banana        6032 393.350000
             production profits
year fruit
2001 apple         2345    245.6
     banana        5632    432.7
2002 apple         3245    534.1
     banana        6432    354.0
2003 apple         4532    467.8