（六）pandas 日常使用技巧

pandas数据处理

1、删除重复元素

import numpy as np
import pandas  as pd
from pandas import Series,DataFrame

df = DataFrame({"color":["red","white","red","green"], "size":[10,20,10,30]})
df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size
0	red	10
1	white	20
2	red	10
3	green	30

使用duplicated()函数检测重复的行，返回元素为布尔类型的Series对象，每个元素对应一行，如果该行不是第一次出现，则元素为True

df.duplicated()

0    False
1    False
2     True
3    False
dtype: bool

使用drop_duplicates()函数删除重复的行

df.drop_duplicates()

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size
0	red	10
1	white	20
3	green	30

如果使用pd.concat([df1,df2],axis = 1)生成新的DataFrame，新的df中columns相同，使用duplicate()和drop_duplicates()不会出错！！！！！

df2  =pd.concat((df,df),axis = 1)
df2

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size	color	size
0	red	10	red	10
1	white	20	white	20
2	red	10	red	10
3	green	30	green	30

df2.duplicated()

0    False
1    False
2     True
3    False
dtype: bool

df2.drop_duplicates()

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size	color	size
0	red	10	red	10
1	white	20	white	20
3	green	30	green	30

2. 映射

映射的含义：创建一个映射关系列表，把values元素和一个特定的标签或者字符串绑定

需要使用字典：

map = { 'label1':'value1', 'label2':'value2', ... }

包含三种操作：

• replace()函数：替换元素
• 最重要：map()函数：新建一列
• rename()函数：替换索引

1) replace()函数：替换元素

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size
0	red	10
1	white	20
2	red	10
3	green	30

使用replace()函数，对values进行替换操作

首先定义一个字典

color = {"red":10,"green":20}

调用.replace()

df.replace(color, inplace=True)

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size
0	10	10
1	white	20
2	10	10
3	20	30

replace还经常用来替换NaN元素

df.loc[1] = np.nan

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size
0	10	10.0
1	NaN	NaN
2	10	10.0
3	20	30.0

v = {np.nan:0.1}
df.replace(v)

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	size
0	10.0	10.0
1	0.1	0.1
2	10.0	10.0
3	20.0	30.0

============================================

练习19：

假设张三李四的课表里有满分的情况，老师认为是作弊，把所有满分的情况（包括150,300分）都记0分，如何实现？

============================================

2) map()函数：新建一列

使用map()函数，由已有的列生成一个新列

适合处理某一单独的列。

df = DataFrame(np.random.randint(0,150,size  =(4,4)),columns = ["Python","Java","PHP","HTML"],
               index = ["张三","旭日","阳刚","木兰"])
df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML
张三	72	33	12	62
旭日	128	12	92	127
阳刚	133	54	89	31
木兰	90	144	136	118

仍然是新建一个字典

v = {72:90,128:100,133:134,90:43}

df["Go"] = df["Python"].map(v)

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go
张三	72	33	12	62	90
旭日	128	12	92	127	100
阳刚	133	54	89	31	134
木兰	90	144	136	118	43

map()函数中可以使用lambda函数

df["C"] = df["Go"].map(lambda x : x - 40)

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C
张三	72	33	12	62	90	50
旭日	128	12	92	127	100	60
阳刚	133	54	89	31	134	94
木兰	90	144	136	118	43	3

def mp(x):
    if x < 50:
        return "不及格"
    else:
        return "优秀"
#添加回函数 这个时候可来更加复杂的操作

df["score"] = df['C'].map(mp)

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C	score
张三	72	33	12	62	90	50	优秀
旭日	128	12	92	127	100	60	优秀
阳刚	133	54	89	31	134	94	优秀
木兰	90	144	136	118	43	3	不及格

transform()和map()类似

df["score2"] = df["C"].transform(mp)

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C	score	score2
张三	72	33	12	62	90	360	优秀	优秀
旭日	128	12	92	127	100	400	优秀	优秀
阳刚	133	54	89	31	134	536	优秀	优秀
木兰	90	144	136	118	43	172	不及格	不及格

使用map()函数新建一个新列

#可以不可以修改当前的列？？？
df["C"] = df["C"].map(lambda x : x*2)

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C	score	score2
张三	72	33	12	62	90	360	优秀	优秀
旭日	128	12	92	127	100	400	优秀	优秀
阳刚	133	54	89	31	134	536	优秀	优秀
木兰	90	144	136	118	43	172	不及格	不及格

============================================

练习20：

新增两列，分别为张三、李四的成绩状态，如果分数低于90，则为"failed"，如果分数高于120，则为"excellent"，其他则为"pass"

【提示】使用函数作为map的参数

============================================

3) rename()函数：替换索引

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C	score	score2
张三	72	33	12	62	90	720	优秀	优秀
旭日	128	12	92	127	100	800	优秀	优秀
阳刚	133	54	89	31	134	1072	优秀	优秀
木兰	90	144	136	118	43	344	不及格	不及格

def cols(x):
    if x == "Python":
        return "大蟒蛇"
    if x == "PHP":
        return "php"
    else:
        return x

inds = {'张三':"Zhang sir", '木兰':"Miss hua"}

仍然是新建一个字典

使用rename()函数替换行索引

df.rename(index = inds, columns = cols)

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	大蟒蛇	Java	php	HTML	Go	C	score	score2
Zhang sir	72	33	12	62	90	720	优秀	优秀
旭日	128	12	92	127	100	800	优秀	优秀
阳刚	133	54	89	31	134	1072	优秀	优秀
Miss hua	90	144	136	118	43	344	不及格	不及格

3. 异常值检测和过滤

使用describe()函数查看每一列的描述性统计量

df.describe()

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C
count	4.00000	4.000000	4.000000	4.000000	4.000000	4.000000
mean	105.75000	60.750000	82.250000	84.500000	91.750000	734.000000
std	29.57899	58.088295	51.525883	45.814845	37.562171	300.497365
min	72.00000	12.000000	12.000000	31.000000	43.000000	344.000000
25%	85.50000	27.750000	69.750000	54.250000	78.250000	626.000000
50%	109.00000	43.500000	90.500000	90.000000	95.000000	760.000000
75%	129.25000	76.500000	103.000000	120.250000	108.500000	868.000000
max	133.00000	144.000000	136.000000	127.000000	134.000000	1072.000000

使用std()函数可以求得DataFrame对象每一列的标准差

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C	score	score2
张三	72	33	12	62	90	720	优秀	优秀
旭日	128	12	92	127	100	800	优秀	优秀
阳刚	133	54	89	31	134	1072	优秀	优秀
木兰	90	144	136	118	43	344	不及格	不及格

df.std()

Python     29.578990
Java       58.088295
PHP        51.525883
HTML       45.814845
Go         37.562171
C         300.497365
dtype: float64

根据每一列的标准差，对DataFrame元素进行过滤。

借助any()函数, 测试是否有True，有一个或以上返回True，反之返回False

对每一列应用筛选条件,去除标准差太大的数据

df.drop(["score","score2"], axis = 1, inplace=True)

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	Python	Java	PHP	HTML	Go	C
张三	72	33	12	62	90	720
旭日	128	12	92	127	100	800
阳刚	133	54	89	31	134	1072
木兰	90	144	136	118	43	344

df2 = df.stack().unstack(level = 0)
df2

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	张三	旭日	阳刚	木兰
Python	72	128	133	90
Java	33	12	54	144
PHP	12	92	89	136
HTML	62	127	31	118
Go	90	100	134	43
C	720	800	1072	344

import numpy as np

cond = np.abs(df2) < df2.std()*2
cond

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	张三	旭日	阳刚	木兰
Python	True	True	True	True
Java	True	True	True	True
PHP	True	True	True	True
HTML	True	True	True	True
Go	True	True	True	True
C	False	False	False	False

df2.std()

张三    273.401109
旭日    292.212537
阳刚    403.757064
木兰    103.765922
dtype: float64

df.std(axis = 1)

张三    273.401109
旭日    292.212537
阳刚    403.757064
木兰    103.765922
dtype: float64

df2[cond].dropna()

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	张三	旭日	阳刚	木兰
Python	72.0	128.0	133.0	90.0
Java	33.0	12.0	54.0	144.0
PHP	12.0	92.0	89.0	136.0
HTML	62.0	127.0	31.0	118.0
Go	90.0	100.0	134.0	43.0

删除特定索引df.drop(labels,inplace = True)

============================================

练习21：

新建一个形状为10000*3的标准正态分布的DataFrame(np.random.randn)，去除掉所有满足以下情况的行：其中任一元素绝对值大于3倍标准差

============================================

n = np.random.randn(10000,3)

df = DataFrame(n)
df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	0	1	2
0	2.881592	0.536820	1.216572
1	0.456766	1.395878	1.768264
2	1.708867	1.622249	0.335690
3	0.194154	-0.487591	-0.015412
4	-2.255826	-0.237842	1.419884
5	1.049099	-0.366917	-0.042190
6	0.191674	-2.372953	1.019347
7	-0.838643	-0.399063	-1.339320
8	1.517263	-0.761005	-1.950791
9	0.251293	0.691856	-0.434976
10	-0.393337	-0.840542	1.051823
11	0.519547	-0.960125	0.693721
12	0.675356	0.742952	-1.987214
13	-1.073620	-1.786886	0.286581
14	-1.137472	-1.294179	-1.650784
15	-0.211439	-0.398124	-0.564845
16	0.150546	-0.094917	0.389879
17	0.202585	0.345154	-0.579804
18	-0.591010	-0.963711	1.492271
19	1.184359	-0.888860	-0.377440
20	-1.122213	0.263416	1.482226
21	-0.181044	-0.890953	1.385926
22	-1.860743	1.028910	0.016576
23	1.289668	0.079026	0.391087
24	-1.501513	1.269525	0.026053
25	-0.845240	0.744394	-1.451082
26	-1.094819	-0.503675	0.403650
27	1.037809	-0.475193	-1.582079
28	-1.655881	-0.532378	0.668746
29	2.176592	-1.564236	1.892409
...	...	...	...
9970	0.163431	-0.453160	-0.551507
9971	-1.818862	-0.315904	0.254854
9972	-0.284665	0.446465	-0.095406
9973	1.951441	0.167062	1.005489
9974	-0.139046	0.300747	-0.964243
9975	-0.292296	0.733086	1.749265
9976	0.565221	0.365676	0.724422
9977	0.554723	1.523374	-2.181834
9978	-1.321702	-2.075783	0.570540
9979	0.619274	-0.393143	-0.809066
9980	0.879297	-0.476391	-0.004182
9981	1.230847	0.951403	2.314687
9982	0.645433	0.313307	0.831975
9983	-0.317260	-0.246456	0.704056
9984	-0.698464	0.002091	0.498848
9985	0.593881	1.192555	-0.025465
9986	-1.343395	-1.148288	0.153664
9987	1.442074	-1.500158	-0.105832
9988	0.767976	0.209889	-0.486307
9989	0.832209	-0.969938	-0.664690
9990	-0.872977	0.166470	-0.534711
9991	-1.368020	-0.477498	0.157921
9992	0.449316	-0.021680	0.109007
9993	-0.967712	1.411765	0.529959
9994	-0.007388	0.807077	0.295686
9995	-0.241627	0.256662	0.890862
9996	-0.082404	1.090093	0.180587
9997	-1.086674	0.879875	-1.547565
9998	-0.639018	0.176242	-0.230805
9999	0.487361	-0.096955	0.262908

10000 rows × 3 columns

cond = np.abs(df) > df.std()*3
cond

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	0	1	2
0	False	False	False
1	False	False	False
2	False	False	False
3	False	False	False
4	False	False	False
5	False	False	False
6	False	False	False
7	False	False	False
8	False	False	False
9	False	False	False
10	False	False	False
11	False	False	False
12	False	False	False
13	False	False	False
14	False	False	False
15	False	False	False
16	False	False	False
17	False	False	False
18	False	False	False
19	False	False	False
20	False	False	False
21	False	False	False
22	False	False	False
23	False	False	False
24	False	False	False
25	False	False	False
26	False	False	False
27	False	False	False
28	False	False	False
29	False	False	False
...	...	...	...
9970	False	False	False
9971	False	False	False
9972	False	False	False
9973	False	False	False
9974	False	False	False
9975	False	False	False
9976	False	False	False
9977	False	False	False
9978	False	False	False
9979	False	False	False
9980	False	False	False
9981	False	False	False
9982	False	False	False
9983	False	False	False
9984	False	False	False
9985	False	False	False
9986	False	False	False
9987	False	False	False
9988	False	False	False
9989	False	False	False
9990	False	False	False
9991	False	False	False
9992	False	False	False
9993	False	False	False
9994	False	False	False
9995	False	False	False
9996	False	False	False
9997	False	False	False
9998	False	False	False
9999	False	False	False

10000 rows × 3 columns

drop_index = df[cond.any(axis = 1)].index

df2 = df.drop(drop_index)

df2.shape

(9927, 3)

4. 排序

使用.take()函数排序

可以借助np.random.permutation()函数随机排序

df = DataFrame(np.random.randint(0,150,size = (4,4)), columns = ["python","java","php","html"],
               index = ["张三","旭日","阳刚","木兰"])
df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	python	java	php	html
张三	20	109	25	43
旭日	83	98	22	39
阳刚	144	19	139	131
木兰	142	72	11	103

df.take([3,2,0])

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	python	java	php	html
木兰	142	72	11	103
阳刚	144	19	139	131
张三	20	109	25	43

indices = np.random.permutation(3)

indices

array([2, 0, 1])

df.take(indices)

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	python	java	php	html
阳刚	144	19	139	131
张三	20	109	25	43
旭日	83	98	22	39

随机抽样

当DataFrame规模足够大时，直接使用np.random.randint()函数，就配合take()函数实现随机抽样

df2 = DataFrame(np.random.randn(10000,3))
df2

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	0	1	2
0	-0.025224	-0.565157	0.844056
1	-0.035657	-0.402001	0.472989
2	0.699627	0.803896	0.501812
3	0.245390	1.377563	1.070162
4	-0.748127	-1.863395	-1.011189
5	1.064520	0.941913	1.040098
6	0.342517	-0.420390	0.105190
7	-1.337581	1.902223	-1.237730
8	0.960661	0.510905	-0.702202
9	0.228292	-1.237225	-0.725750
10	0.894908	0.255933	1.285206
11	-0.112649	0.073029	0.226987
12	0.847398	1.278539	-0.316305
13	0.709176	-0.054754	-0.626551
14	-1.492717	-0.270664	1.138691
15	-1.050701	0.731788	1.515430
16	0.033859	-0.481181	0.449713
17	1.908899	0.013049	1.168205
18	2.003074	-0.694794	-2.443718
19	-0.305153	1.659422	-2.338938
20	-0.595257	0.649238	-0.782337
21	-0.143291	-0.661235	-1.292414
22	-0.451794	0.380953	1.187246
23	0.258405	0.352720	-0.671535
24	-1.710904	-1.020995	1.160462
25	-0.790192	0.688780	0.088410
26	-0.174850	-1.112568	-1.633942
27	0.213165	1.020418	0.533577
28	-0.853166	0.192139	2.363981
29	-0.197083	0.637195	-1.911048
...	...	...	...
9970	-1.317070	0.660991	0.393611
9971	-0.947604	-1.415052	1.662456
9972	0.517894	0.179165	0.489423
9973	-0.189215	0.657269	0.047626
9974	1.126898	-0.085763	-1.709755
9975	0.359945	0.411918	0.668606
9976	-0.491320	-1.247942	0.887130
9977	0.736900	0.136471	0.079652
9978	1.469600	0.852718	-0.141616
9979	1.110100	-0.394567	0.997196
9980	-0.581172	-1.658739	1.657382
9981	-1.173605	1.491162	-0.760518
9982	0.097367	0.252979	-1.697217
9983	0.079267	-1.369900	-0.870134
9984	-0.376669	-0.583582	0.250551
9985	0.419189	-0.367227	-0.496057
9986	-0.140032	0.202857	-0.476418
9987	-0.227373	-0.463283	0.559428
9988	-1.595745	0.392217	-0.160671
9989	0.007461	0.840525	0.841650
9990	1.266712	-1.190441	-0.983106
9991	-1.641171	-0.463228	-0.572552
9992	-1.494818	-0.851275	-0.443659
9993	-0.106178	-0.199535	1.542675
9994	-0.433710	-0.561674	-2.116589
9995	0.776234	-1.814600	0.539298
9996	0.099580	-0.133758	1.239752
9997	0.165359	1.558473	0.135779
9998	-0.870957	1.140052	0.056586
9999	-0.390214	-0.152384	-1.184713

10000 rows × 3 columns

indices = np.random.randint(0,10000,size  =10)
df2.take(indices)

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	0	1	2
7761	-0.571778	0.099853	0.026818
1380	0.131348	0.017664	0.983131
733	0.986849	0.262630	-0.551597
2094	-0.514843	0.735007	-1.217740
3761	-1.863511	0.421299	-0.082948
4777	2.176549	1.485876	-0.087476
5176	1.748004	0.498117	1.088707
9606	-1.106140	-1.356788	-1.098564
4736	-1.377561	-0.461284	-1.994532
7965	0.588313	0.024674	0.059207

============================================

练习22：

假设有张三李四王老五的期中考试成绩ddd2，对着三名同学随机排序

============================================

5. 数据聚合

数据聚合是数据处理的最后一步，通常是要使每一个数组生成一个单一的数值。

数据分类处理：

• 分组：先把数据分为几组
• 用函数处理：为不同组的数据应用不同的函数以转换数据
• 合并：把不同组得到的结果合并起来

数据分类处理的核心：

groupby()函数

df.std()

如果想使用color列索引，计算price1的均值，可以先获取到price1列，然后再调用groupby函数，用参数指定color这一列

df = DataFrame({'color':["red","white","red","cyan","cyan","green","white","cyan"],
                "price":np.random.randint(0,8,size  =8),
               "weight":np.random.randint(50,55,size = 8)}
              )
df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	price	weight
0	red	2	53
1	white	5	51
2	red	5	54
3	cyan	5	53
4	cyan	7	52
5	green	5	54
6	white	6	50
7	cyan	0	54

使用.groups属性查看各行的分组情况：

df_sum_price = df.groupby(["color"])[["price"]].sum()
df_sum_price

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	price
color
cyan	12
green	5
red	7
white	11

df_mean_weight  = df.groupby(["color"])[["weight"]].mean()
df_mean_weight

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	weight
color
cyan	53.0
green	54.0
red	53.5
white	50.5

#级联
pd.concat([df,df_sum_price],axis = 1)

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	price	weight	price
0	red	2.0	53.0	NaN
1	white	5.0	51.0	NaN
2	red	5.0	54.0	NaN
3	cyan	5.0	53.0	NaN
4	cyan	7.0	52.0	NaN
5	green	5.0	54.0	NaN
6	white	6.0	50.0	NaN
7	cyan	0.0	54.0	NaN
cyan	NaN	NaN	NaN	12.0
green	NaN	NaN	NaN	5.0
red	NaN	NaN	NaN	7.0
white	NaN	NaN	NaN	11.0

df_mean_weight

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	weight
color
cyan	53.0
green	54.0
red	53.5
white	50.5

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	price	weight
0	red	2	53
1	white	5	51
2	red	5	54
3	cyan	5	53
4	cyan	7	52
5	green	5	54
6	white	6	50
7	cyan	0	54

df_sum = df.merge(df_mean_weight, how = "outer")
df_sum

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	price	weight
0	red	2.0	53.0
1	cyan	5.0	53.0
2	white	5.0	51.0
3	red	5.0	54.0
4	green	5.0	54.0
5	cyan	0.0	54.0
6	cyan	7.0	52.0
7	white	6.0	50.0
8	NaN	NaN	53.5
9	NaN	NaN	50.5

============================================

练习23：

假设菜市场张大妈在卖菜，有以下属性：

菜品(item)：萝卜，白菜，辣椒，冬瓜

颜色(color)：白，青，红

重量(weight)

价格(price)

1. 要求以属性作为列索引，新建一个ddd
2. 对ddd进行聚合操作，求出颜色为白色的价格总和
3. 对ddd进行聚合操作，求出萝卜的所有重量(包括白萝卜，胡萝卜，青萝卜）以及平均价格
4. 使用merge合并总重量及平均价格

============================================

6.0 高级数据聚合

可以使用pd.merge()函数将聚合操作的计算结果添加到df的每一行

使用groupby分组后调用加和等函数进行运算，让后最后可以调用add_prefix()，来修改列名

可以使用transform和apply实现相同功能

在transform或者apply中传入函数即可

df["columns"] = df["color"].map(sum)

---------------------------------------------------------------------------

TypeError                                 Traceback (most recent call last)

<ipython-input-132-6a8b5973654d> in <module>()
----> 1 df["columns"] = df["color"].map(sum)
      2 df

C:\anaconda\lib\site-packages\pandas\core\series.py in map(self, arg, na_action)
   2352         else:
   2353             # arg is a function
-> 2354             new_values = map_f(values, arg)
   2355
   2356         return self._constructor(new_values,

pandas/_libs/src/inference.pyx in pandas._libs.lib.map_infer()

TypeError: unsupported operand type(s) for +: 'int' and 'str'

sum([10,2])

12

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	price	weight
0	red	2	53
1	white	5	51
2	red	5	54
3	cyan	5	53
4	cyan	7	52
5	green	5	54
6	white	6	50
7	cyan	0	54

df.groupby("color")[["price","weight"]].transform(sum)

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	price	weight
0	7	107
1	11	101
2	7	107
3	12	159
4	12	159
5	5	54
6	11	101
7	12	159

df

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	color	price	weight
0	red	2	53
1	white	5	51
2	red	5	54
3	cyan	5	53
4	cyan	7	52
5	green	5	54
6	white	6	50
7	cyan	0	54

df.groupby("color")[["price","weight"]].apply(sum)

.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

	price	weight
color
cyan	12	159
green	5	54
red	7	107
white	11	101

transform()与apply()函数还能传入一个函数或者lambda

df = DataFrame({'color':['white','black','white','white','black','black'], 'status':['up','up','down','down','down','up'], 'value1':[12.33,14.55,22.34,27.84,23.40,18.33], 'value2':[11.23,31.80,29.99,31.18,18.25,22.44]})

apply的操作对象，也就是传给lambda的参数是整列的数组

============================================

练习24：

使用transform与apply实现练习23的功能

============================================