1
 #分析网页后可以得知get历史所有数据的参数
2
url='https://datachart.500.com/ssq/history/newinc/history.php?start=03001'
3

4
#加载相关的库
5
import requests
6
import numpy as np
7
import pandas as pd
8

9
#获取历史所有双色球数据
10
response = requests.get(url)
11
response.encoding = 'utf-8'
12
re_text = response.text
13

14
#网页数据解析
15
re=re_text.split('<tbody id="tdata">')[1].split('</tbody>')[0]
16
result=re.split('<tr class="t_tr1">')[1:]
17

18
all_numbers=[]
19
for i in result:
20
    each_numbers=[]
21
    i=i.replace('<!--<td>2</td>-->','')
22
    each=i.split('</td>')[:-1]
23
    for j in each:
24
        each_numbers.append(j.split('>')[1].replace('&nbsp;',''))
25

26
    all_numbers.append(each_numbers)
27

28
#定义列名称
29
col=['期号','红球1','红球2','红球3','红球4','红球5','红球6','蓝球','快乐星期天','奖池奖金(元)',
30
     '一等奖注数','一等奖奖金(元)','二等奖注数','二等奖奖金(元)','总投注额(元)','开奖日期']
31

32
#解析完网页数据，生成双色球数据框
33
df_all=pd.DataFrame(all_numbers,columns=col)
34
df_all.head()

1
#日期转换
2
df_all['开奖日期_dt']=pd.to_datetime(df_all['开奖日期'])
3
df_all['year']=df_all['开奖日期_dt'].dt.year
4
df_all['month']=df_all['开奖日期_dt'].dt.month
5
df_all['day']=df_all['开奖日期_dt'].dt.day
6
df_all['weekday']=df_all['开奖日期_dt'].dt.weekday_name
7
df_all.head()

1
#one-hot 编码转换自定义函数
2
def lotterydata(df):
3
    modeldata=df.copy()
4

5
    redball=[]
6
    for i in range(1,34):
7
        redball.append('红球'+'%02d'%i)
8
    for i in redball:
9
        modeldata[i]=0
10

11
    blueball=[]
12
    for i in range(1,17):
13
        blueball.append('蓝球'+'%02d'%i)
14
    for i in blueball:
15
        modeldata[i]=0
16

17

18
    for row in range(modeldata.shape[0]):
19
        #print(row)
20
        #print(modeldata.iloc[row,:])
21
        for i in redball:
22
            #print(i)
23
            #modeldata[i]=0
24
            if (modeldata.iloc[row,:]['红球1']==i[-2:] or modeldata.iloc[row,:]['红球2']==i[-2:]
25
                or modeldata.iloc[row,:]['红球3']==i[-2:] or modeldata.iloc[row,:]['红球4']==i[-2:]
26
                or modeldata.iloc[row,:]['红球5']==i[-2:] or modeldata.iloc[row,:]['红球6']==i[-2:]):
27
                modeldata.loc[row,i]=1
28

29
        for j in blueball:
30
            #modeldata[j]=0
31
            if modeldata.iloc[row,:]['蓝球']==j[-2:]:
32
                modeldata.loc[row,j]=1
33
    return modeldata
34

35
#生成各颜色球的0-1编码
36
modeldata=lotterydata(df_all)
37
modeldata.head()

1
allhistorydata=modeldata.iloc[:,-49:].copy()
2

3
#历史所有红球和蓝球数据
4
allhistorydata_red=allhistorydata.iloc[:,:33]
5
allhistorydata_blue=allhistorydata.iloc[:,-16:]
6

7
#最近20期红球和最近48期蓝球
8
#（33*3）/6  每个红球有3次出现机会，看一共需要多少期，这里取整数20期
9
#（16*3）/1  每个蓝球有3次出现机会，看一共需要多少期
10
recently20_red=allhistorydata.iloc[:20,:33]
11
recently48_blue=allhistorydata.iloc[:48,-16:]
12

13
#求和
14
historyred_sum=allhistorydata_red.sum()
15
historyblue_sum=allhistorydata_blue.sum()
16

17
recently20red_sum=recently20_red.sum()
18
recently48blue_sum=recently48_blue.sum()
19

20
#排序
21
historyred_sum=historyred_sum.sort_values(ascending=True)
22
historyblue_sum=historyblue_sum.sort_values(ascending=True)
23

24
recently20red_sum=recently20red_sum.sort_values(ascending=True)
25
recently48blue_sum=recently48blue_sum.sort_values(ascending=True)
26

27
#数据展示
28
import matplotlib.pyplot as plt
29

30
%matplotlib inline
31
plt.rcParams['font.sans-serif'] = ['SimHei']    #显示中文
32

33
plt.figure(figsize=(30,24),facecolor='snow')
34

35
#历史出现次数最少的10个红球
36
x_red=historyred_sum.index.map(lambda x:x[-2:])[:10]
37
y_red=historyred_sum.values[:10]
38

39
#历史出现次数最少的5个蓝球
40
x_blue=historyblue_sum.index.map(lambda x:x[-2:])[:5]
41
y_blue=historyblue_sum.values[:5]
42

43
plt.subplot(3,2,1)
44
plt.bar(x_red,y_red,width=0.4,align='center',color='r')
45
for a,b in zip(x_red,y_red):
46
    plt.text(a,b,b,ha='center',va='bottom',fontsize=15)
47
plt.tick_params(axis='x',labelsize=30)
48
plt.title("历史出现次数最少的10个红球",fontsize=30)
49

50
plt.subplot(3,2,2)
51
plt.bar(x_blue,y_blue,width=0.2,align='center',color='b')
52
for a,b in zip(x_blue,y_blue):
53
    plt.text(a,b,b,ha='center',va='bottom',fontsize=15)
54
plt.tick_params(axis='x',labelsize=30)
55
plt.title("历史出现次数最少的5个蓝球",fontsize=30)
56

57
#最近20期红球
58
x20_red=recently20red_sum.index.map(lambda x:x[-2:])
59
y20_red=recently20red_sum.values
60

61
#最近48期蓝球
62
x48_blue=recently48blue_sum.index.map(lambda x:x[-2:])
63
y48_blue=recently48blue_sum.values
64

65
plt.subplot(3,1,2)
66
plt.bar(x20_red,y20_red,width=0.5,align='center',color='r')
67
for a,b in zip(x20_red,y20_red):
68
    plt.text(a,b,b,ha='center',va='bottom',fontsize=15)
69
plt.tick_params(axis='x',labelsize=25)
70
plt.title("最近20期红球情况",fontsize=30)
71

72
plt.subplot(3,1,3)
73
plt.bar(x48_blue,y48_blue,width=0.5,align='center',color='b')
74
for a,b in zip(x20_blue,y20_blue):
75
    plt.text(a,b,b,ha='center',va='bottom',fontsize=15)
76
plt.tick_params(axis='x',labelsize=25)
77
plt.title("最近48期蓝球情况",fontsize=30)
78

79
plt.show()