资讯详情

DOTA2利雅得大师赛利用api多线程对选手数据和战队数据爬取与分析

首先，恭喜中国队LGD赢得利雅得大师赛冠军！

数据的爬取

选手数据和团队数据爬行

数据分析

数据(因为原数据的英雄都是id，将英雄连接到这里id用英雄名代替）

选手KDA前十名

参战率、参葬率、战死率

场均十分钟补刀前十，以及他们GPM,XPM

场均反眼数前十的选手

队伍场均治疗量

编辑

英雄在这场比赛中的前十名和胜率

计算先摧毁对方优势路、中路、劣势路一塔的胜率

最后看看LGD英雄和胜率的使用

数据的爬取

数据来源于opendota，对opendota的api调用获取数据，找到利雅得的所有比赛id，那就要先找到这个比赛，首先咱们来看看opendota的api页面，网站OpenDota API

1043127f363546a592d4fc3dab3b1f0b.png

打开到match页面将看到请求数据api介绍及对应url里面的json数据

我们在利雅得大师赛中找到一场比赛，比如6676393091

利用api提供的网站打开看看

一大串json这样看数据不方便，按F12打开抓包工具，看看里面

可以看出这场比赛的数据有很多分组，相应的分组是什么意思？api查看文档，其中league中的leagueid这是利雅得大师赛的比赛id为14391，这就是dota2不同比赛之间的区别，然后通过leagueid获取到利雅得大师赛的所有比赛id，然后进行数据爬取，下面我直接上代码

首先是导入所需的模块，爬上利雅得的所有比赛id，参数id是利雅得大师赛id 14391

import os.path import time import pandas as pd import requests import threading   def pro_game(id):     pro_match_id = []     pro_url = f'https://api.opendota.com/api/leagues/{id}/matches'     text = requests.get(url=pro_url,headers=headers).json()     for item in text:         pro_match_id.append(item['match_id'])     print(pro_match_id)     return pro_match_id

获取的比赛id之后，我们需要爬取每场比赛的数据。爬行前，我们需要比赛id进行post请求，因为如果执行post的话，opendota内置功能将进一步分析视频，从而获得更多的数据，如推塔时间团战爆发次数等get得不到的，一定要先post请求一下

def jiexi(match_id):     post_url = f'https://api.opendota.com/api/request/{match_id}'     pos = requests.post(url=post_url, headers=headers)     match_url = f'https://api.opendota.com/api/matches/{match_id}'     match_text = requests.get(url=match_url, headers=headers).json()     return match_text

get在获得每场比赛的数据后，您可以选择您想要的数据进行爬行，选手的数据爬行和存储如下：

本来想看看能不能定位选手是哪条路，但是因为dota2.玩法多样，玩家每场分路也不一样，导致分路数据不太准确，所以我在后面的数据分析中没有用到他

def role(lr):     if lr == 1:         return '优势路'     if lr == 2:         return '中路'     if lr == 3:         return '劣势路'

根据选手的哪一方和哪一方的胜利来判断选手是否赢得了这场比赛

选手数据和团队数据爬行

def xuanshou(match_text):     print(开始)     for item in match_text['players']:         row_dic = {}         try:             row_dic['match_id'] = item['match_id']             row_dic['name'] = item['name']             row_dic['ID'] = item['account_id']             row_dic['英雄id'] = item['hero_id']             l_role = item['lane_role']             row_dic[位置] = role(l_role)             row_dic[10分钟补刀] = item['benchmarks']['lhten']['raw']             row_dic[杀人数] = item['kills']             row_dic[死亡数] = item['deaths']             row_dic[助攻] = item['assists']             row_dic[反眼数] = item['observer_kills']             row_dic[总金钱] = item['total_gold']             row_dic['治疗量'] = item['hero_healing']             row_dic['GPM'] = item['benchmarks']['gold_per_min']['raw']             row_dic['XPM'] = item['benchmarks']['xp_per_min']['raw']             row_dic[每分钟补刀数] = item['benchmarks']['last_hits_per_min']['raw']             row_dic[每分钟伤害] = item['benchmarks']['hero_damage_per_min']['raw']             row_dic[总伤害] = item['hero_damage']             row_dic[] = match_text['radiant_team']['name']             row_dic[天辉杀人数] = match_text['radiant_score']             row_dic[] = match_text['dire_team']['name']             row_dic[] = match_text['dire_score']             place = item['isRadiant']              if place == True:                 row_dic[所在方] = '天辉'                 if item['win'] == 0:                     row_dic[输赢] = '输'                 if item['win'] == 1:                     row_dic[输赢] = '赢'             if place == False:                 row_dic[所在方] = '夜宴'                 if item['win'] == 0:                     row_dic[输赢] = '输'                 if item['win'] == 1:                     row_dic[输赢] = '赢'         except:             print(‘出错’)         row_list.append(row_dic)       time.sleep(0.5)
def paqu(match_text):     dicq = {}    dicq['比赛id'] = match_text['match_id']
    dicq['持续时间'] = match_text['duration']/60
    dicq['天辉方战队'] = match_text['radiant_team']['name']
    dicq['天辉杀人数'] = match_text['radiant_score']
    dicq['夜宴方战队'] = match_text['dire_team']['name']
    dicq['夜宴杀人数'] = match_text['dire_score']
    dicq['胜利方'] = '天辉' if match_text['radiant_win'] == True else '夜宴'
    dicq['团战次数'] = len(match_text['teamfights'])
    for item in match_text['objectives']:
        if item['type']=='building_kill' and item['key'] =='npc_dota_goodguys_tower1_mid':
            dicq['天辉中路一塔'] = item['time']/60
        if item['type']=='building_kill' and item['key'] =='npc_dota_goodguys_tower1_top':
            dicq['天辉劣势路一塔'] = item['time']/60
        if item['type']=='building_kill' and item['key'] =='npc_dota_goodguys_tower1_bot':
            dicq['天辉优势路一塔'] = item['time']/60
        if item['type'] == 'building_kill' and item['key'] =='npc_dota_badguys_tower1_mid':
            dicq['夜宴中路1塔'] = item['time'] / 60
        if item['type']=='building_kill' and item['key'] =='npc_dota_badguys_tower1_top':
            dicq['夜宴优势路一塔'] = item['time']/60
        if item['type']=='building_kill' and item['key'] =='npc_dota_badguys_tower1_bot':
            dicq['夜宴劣势路一塔'] = item['time']/60
    quanju_list.append(dicq)
    print(f"{dicq['天辉方战队']}vs{dicq['夜宴方战队']}完成")

主函数里面使用双线程进行数据的爬取 

if __name__ =='__main__':
    # account_id = 1150728771
    # m = match(account_id,100)
    where = input('请输入文件存储地址:')
    name = 'DOTA2利雅得.xlsx'
    dizhi = os.path.join(where,name)
    id = pro_game(14391)
    print(id)
    row_list = []
    quanju_list = []
    for mid in id:
        shuju = jiexi(mid)
        thread1 = threading.Thread(target=xuanshou,args=(shuju,))
        thread2 = threading.Thread(target=paqu,args=(shuju,))
        thread1.start()
        thread2.start()
    zd_d = pd.DataFrame(quanju_list)
    r_d = pd.DataFrame(row_list)
    r_d.to_excel(dizhi,sheet_name='选手数据')
    with pd.ExcelWriter(dizhi,mode='a',engine='openpyxl')as writer:
        zd_d.to_excel(writer,sheet_name='战队数据')

运行程序后结果大概是这样

 

 数据分析

import pandas as pd 
import matplotlib.pyplot as plt
lujing = r"E:/python文件/DOTA2利雅得.xlsx"
lujing1 = r"E:\python文件\DOTA2数据.xlsx"
shuju = pd.read_excel(lujing,sheet_name='选手数据')
shuju1 = pd.read_excel(lujing1,sheet_name='dota2英雄id')
shuju = pd.merge(shuju,shuju1,on='英雄id')
shuju.head()

数据（因为原数据的英雄都是id，这里进行连接将英雄id替换为英雄名 ）

 选手KDA前十名

def kda(shuju):
    if shuju['死亡数'] == 0:     
        k = shuju['杀人数']+shuju['助攻']
        return k
    else:
        k = (shuju['杀人数']+shuju['助攻'])/shuju['死亡数']
        return k 
shuju['KDA'] = shuju.apply(kda,axis=1)
kda_m = shuju.groupby(['name','战队'])['KDA'].mean()
kda_m.sort_values(ascending=False,inplace=True)
kda_m.head(10)

 选手参战率，参葬率，战死率

def canzhan(shuju):
    if shuju['所在方'] =='天辉':
        x = (shuju['杀人数']+shuju['助攻'])/shuju['天辉杀人数']
        
    else:
        x = (shuju['杀人数']+shuju['助攻'])/shuju['夜宴杀人数']
    return x

def canzang(shuju):
    if shuju['所在方'] =='天辉':
        y = shuju['死亡数']/shuju['夜宴杀人数']
    else:
        y = shuju['死亡数']/shuju['天辉杀人数']
    return y   
shuju['参战率'] = shuju.apply(canzhan,axis=1)
shuju['参葬率'] = shuju.apply(canzang,axis=1)
x = shuju.loc[(shuju['战队']=='PSG.LGD')&(shuju['输赢']=='输'),:]
x = shuju.groupby(['name','战队'])[['参葬率','参战率']].mean()
x['战死率'] = x['参葬率']/x['参战率']
x.sort_values(by='战死率',ascending=True).head(10)

选手场均十分钟补刀数前十,以及他们的GPM,XPM

budao = shuju.groupby(['name','战队'])[['10分钟补刀','GPM','XPM']].mean()
budao = budao.reset_index(drop=False)

budao.sort_values(by='10分钟补刀',ascending=False).head(10)

 

 

选手场均反眼数前十 

fanyan = shuju.groupby(['name','战队'])['反眼数'].mean()
fanyan = fanyan.reset_index()
fy = fanyan.sort_values(by='反眼数',ascending = False)
fy.head(10)

战队场均治疗量

zhiliao = shuju.groupby(['战队'])['治疗量'].mean()
zhiliao.sort_values(ascending=False,inplace=True)
x = zhiliao.reset_index(drop=False)
x

 

 本次比赛英雄的上场数前十以及胜率

changshu = shuju.groupby(['英雄名'])['输赢'].count()
hero_wl = shuju.groupby(['英雄名'])['输赢'].value_counts()
wl = hero_wl[:,'赢']
hero = pd.merge(changshu,wl,on='英雄名')
hero['胜率'] = hero['输赢_y']/hero['输赢_x']
hero.rename(columns={'输赢_x':'场数','输赢_y':'胜场'},inplace = True)
y =hero.sort_values(['场数','胜场'],ascending=False).head(10)
y

 计算先摧毁对方优势路，中路，劣势路一塔的胜率

x.fillna(100,inplace=True)
tiaojian1 = '天辉中路一塔>夜宴中路1塔 and 胜利方=="天辉"'
tiaojian2 = '天辉中路一塔<夜宴中路1塔 and 胜利方=="夜宴"'
y1 = x.query(tiaojian1)
y2 = x.query(tiaojian2)
(len(y1)+len(y2))/len(x)

 0.722

tiaojian1 = '天辉优势路一塔>夜宴优势路一塔 and 胜利方=="天辉"'
tiaojian2 = '天辉优势路一塔<夜宴优势路一塔 and 胜利方=="夜宴"'
y1 = x.query(tiaojian1)
y2 = x.query(tiaojian2)
(len(y1)+len(y2))/len(x)

 

 0.592

tiaojian1 = '天辉劣势路一塔>夜宴劣势路一塔 and 胜利方=="天辉"'
tiaojian2 = '天辉劣势路一塔<夜宴劣势路一塔 and 胜利方=="夜宴"'
y1 = x.query(tiaojian1)
y2 = x.query(tiaojian2)

(len(y1)+len(y2))/len(x)

 0.759

最后看看LGD的使用英雄以及胜率

tiaojian = "战队=='PSG.LGD'"
sec = shuju.query(tiaojian)
changshu = sec.groupby(['英雄名'])['输赢'].count()
win = sec.groupby(['英雄名'])['输赢'].value_counts()
wl = win[:,'赢']
hero = pd.merge(changshu,wl,on='英雄名')
hero['胜率'] = hero['输赢_y']/hero['输赢_x']
hero.rename(columns={'输赢_x':'场数','输赢_y':'胜场'},inplace = True)
y =hero.sort_values(['场数','胜场'],ascending=False).head(10)
y

 

 

 

标签： psg3m磁感应传感器

锐单商城拥有海量元器件数据手册、 IC替代型号，打造 电子元器件IC百科大全！

锐单商城 - 一站式电子元器件采购平台