PyPackage01---Pandas00_DataFrame

pandas DataFrame基础操作，不定时update

创建数据框

构造两个数据集

• df1:用户基础属性，年龄、性别、注册时间
• df2:用户交易属性，交易时间、支付金额

从dict创建

import pandas as pd

userid = list(range(1, 6))
sex = ["male", "female","male", "female","male"]
createtime_str = [
    "2019-01-01 11:45:50", "2019-01-02 11:55:50", "2019-01-21 11:45:50", 
    "2019-02-01 12:45:50", "2019-01-15 10:40:50"
]
buytime_str = ["2019-04-01 11:45:50", "2019-05-02 11:56:50", "2019-07-21 12:45:50", 
    "2019-08-01 12:40:50", "2019-01-06 10:00:50"]
age = [18, 37, 21, 44, 39]
payamount =[11.15, 10.37, 12.11, 14.5, 16.39]

df1 = pd.DataFrame({'userid':userid,'sex':sex,'age':age,'createtime_str':createtime_str})
df2 = pd.DataFrame({'userid':userid,'buytime_str':buytime_str,'payamount':payamount})

df1

	userid	sex	age	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

df2

	userid	buytime_str	payamount
0	1	2019-04-01 11:45:50	11.15
1	2	2019-05-02 11:56:50	10.37
2	3	2019-07-21 12:45:50	12.11
3	4	2019-08-01 12:40:50	14.50
4	5	2019-01-06 10:00:50	16.39

df1

	userid	sex	age	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

数据框概览

判断类型

type(df1)

pandas.core.frame.DataFrame

打印列类型

df1.dtypes

userid             int64
sex               object
age                int64
createtime_str    object
dtype: object

描述性统计

只输出了数值型列的相关信息

df1.describe()

	userid	age
count	5.000000	5.000000
mean	3.000000	31.800000
std	1.581139	11.562872
min	1.000000	18.000000
25%	2.000000	21.000000
50%	3.000000	37.000000
75%	4.000000	39.000000
max	5.000000	44.000000

行数列数统计

df1.shape
print("行数："+str(df1.shape[0])+"\n"+"列数："+str(df1.shape[1]))

行数：5
列数：4

取子集

取某个位置

行列索引从0开始

df1.iloc[1,2]

37

筛选前几行

df1.iloc[1:3,:]

	userid	sex	age	createtime_str
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50

筛选某些列

df1.iloc[:,[1,3]]

	sex	createtime_str
0	male	2019-01-01 11:45:50
1	female	2019-01-02 11:55:50
2	male	2019-01-21 11:45:50
3	female	2019-02-01 12:45:50
4	male	2019-01-15 10:40:50

df1[["userid","age"]]

	userid	age
0	1	18
1	2	37
2	3	21
3	4	44
4	5	39

条件筛选

df1.query("userid>=2")

	userid	sex	age	createtime_str
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

列操作

列数&列名

df1.columns.size

4

list(df1.columns)

['userid', 'sex', 'age', 'createtime_str']

列名修改

修改一个列

# inplace=False标识不覆盖原来的数据框df1
df1.rename(columns={"userid":"id"},inplace=False)

	id	sex	age	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

修改多个列

# inplace=False标识不覆盖原来的数据框df1
df1.rename(columns={"userid":"id","sex":"gender"},inplace=False)

	id	gender	age	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

修改全部列

注意：深拷贝和浅拷贝
https://www.runoob.com/w3cnote/python-understanding-dict-copy-shallow-or-deep.html

import copy
temp = copy.deepcopy(df1)
temp.columns=["id1","sex1","age1","createtime_str"]
temp

	id1	sex1	age1	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

删除列

删除一个列

temp = copy.deepcopy(df1)
temp.drop(axis=1,columns=["sex"],inplace=False)

	userid	age	createtime_str
0	1	18	2019-01-01 11:45:50
1	2	37	2019-01-02 11:55:50
2	3	21	2019-01-21 11:45:50
3	4	44	2019-02-01 12:45:50
4	5	39	2019-01-15 10:40:50

删除多个列

temp = copy.deepcopy(df1)
temp.drop(axis=1,columns=["sex","age"],inplace=False)

	userid	createtime_str
0	1	2019-01-01 11:45:50
1	2	2019-01-02 11:55:50
2	3	2019-01-21 11:45:50
3	4	2019-02-01 12:45:50
4	5	2019-01-15 10:40:50

列筛选

df1[["userid","age"]]

	userid	age
0	1	18
1	2	37
2	3	21
3	4	44
4	5	39

df1.iloc[1:3,0:2]

	userid	sex
1	2	female
2	3	male

增加列

temp = copy.deepcopy(df1)
temp["new"] = temp["age"]*2
temp

	userid	sex	age	createtime_str	new
0	1	male	18	2019-01-01 11:45:50	36
1	2	female	37	2019-01-02 11:55:50	74
2	3	male	21	2019-01-21 11:45:50	42
3	4	female	44	2019-02-01 12:45:50	88
4	5	male	39	2019-01-15 10:40:50	78

增加常数列

temp = copy.deepcopy(df1)
temp["new"] ="new"
temp

	userid	sex	age	createtime_str	new
0	1	male	18	2019-01-01 11:45:50	new
1	2	female	37	2019-01-02 11:55:50	new
2	3	male	21	2019-01-21 11:45:50	new
3	4	female	44	2019-02-01 12:45:50	new
4	5	male	39	2019-01-15 10:40:50	new

通过运算增加列

temp = copy.deepcopy(df1)
temp["new"] = temp.age.apply(lambda x:"20多啦" if x>20 else "还是小伙子哦")
temp

	userid	sex	age	createtime_str	new
0	1	male	18	2019-01-01 11:45:50	还是小伙子哦
1	2	female	37	2019-01-02 11:55:50	20多啦
2	3	male	21	2019-01-21 11:45:50	20多啦
3	4	female	44	2019-02-01 12:45:50	20多啦
4	5	male	39	2019-01-15 10:40:50	20多啦

通过向量or列表增加列

temp = copy.deepcopy(df1)
temp["new"] = list(range(5))
temp

	userid	sex	age	createtime_str	new
0	1	male	18	2019-01-01 11:45:50	0
1	2	female	37	2019-01-02 11:55:50	1
2	3	male	21	2019-01-21 11:45:50	2
3	4	female	44	2019-02-01 12:45:50	3
4	5	male	39	2019-01-15 10:40:50	4

列的类型转化

几个列类型转化

df1.dtypes

userid             int64
sex               object
age                int64
createtime_str    object
dtype: object

temp = copy.deepcopy(df1)
temp["age"] = df1.age.astype("double")
temp.dtypes

userid              int64
sex                object
age               float64
createtime_str     object
dtype: object

多个列批量转化

批量修改目前支持格式不多

• to_numeric
• to_datetime

要么就写循环吧

temp = copy.deepcopy(df1)
temp[["createtime_str"]] = temp[["createtime_str"]].apply(pd.to_datetime,format='%Y-%m-%d %H:%M:%S.%f')
temp.dtypes

userid                     int64
sex                       object
age                        int64
createtime_str    datetime64[ns]
dtype: object

temp

	userid	sex	age	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

列之间的运算

df1.dtypes

userid             int64
sex               object
age                int64
createtime_str    object
dtype: object

import copy
temp = copy.deepcopy(df1)
temp["userid1"] = [5,4,3,2,1]
temp["greatest"] = temp[["userid","userid1"]].apply(lambda x:max(x),axis=1)
temp["leastest"] = temp[["userid","userid1"]].apply(lambda x:min(x),axis=1)
temp

	userid	sex	age	createtime_str	userid1	greatest	leastest
0	1	male	18	2019-01-01 11:45:50	5	5	1
1	2	female	37	2019-01-02 11:55:50	4	4	2
2	3	male	21	2019-01-21 11:45:50	3	3	3
3	4	female	44	2019-02-01 12:45:50	2	4	2
4	5	male	39	2019-01-15 10:40:50	1	5	1

行操作

统计行数

df1.shape[0]

5

行筛选

条件筛选即可

df1.query("userid>=2")

	userid	sex	age	createtime_str
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

重复行删除

import copy
temp = copy.deepcopy(df1)
temp = pd.concat([df1,pd.DataFrame({'userid':[1],'sex':["male"],'age':[18],'createtime_str':["2019-01-01 11:45:50"]})])
temp

	userid	sex	age	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50
0	1	male	18	2019-01-01 11:45:50

temp.drop_duplicates()

	userid	sex	age	createtime_str
0	1	male	18	2019-01-01 11:45:50
1	2	female	37	2019-01-02 11:55:50
2	3	male	21	2019-01-21 11:45:50
3	4	female	44	2019-02-01 12:45:50
4	5	male	39	2019-01-15 10:40:50

                                    2020-02-29 Update于南京市栖霞区