def __init__(self, gm, RoP, main=0):
self.fontsize = 10
self.col_w = 8
self.bt_h = 2
self.bt_w = 10
self.paddingx = 5
self.paddingy = 5
self.RoP = RoP if isinstance(RoP, int) else (1 if 'playoff' in RoP else 0)
self.gm = gm[:-7] if '.pickle' in gm else gm
self.month = int(gm[4:6])
self.season = int(gm[:4]) - 1 if self.month < 9 else int(gm[:4])
game = LoadPickle('D:/sunyiwu/stat/data/seasons_boxscores/%d_%d/%s/%s_boxscores.pickle' %
(self.season, self.season + 1, 'playoff' if self.RoP else 'regular', self.gm))
[self.roadteam, self.hometeam] = [x for x in game[0].keys()]
self.res = game[0]
self.tbs = game[1:]
self.columns = None
if not main:
self.wd_gm = Toplevel()
else:
self.wd_gm = Tk()
self.wd_gm.title('单场比赛')
self.wd_gm.iconbitmap('D:/sunyiwu/stat/images/nbahalfcourt.ico')
self.wd_gm.geometry('+250+100')
self.frame_btn = None
self.trees = [] # 存放主客队各一个数据表格
self.ts = 0
self.pm2pn = LoadPickle('D:/sunyiwu/stat/data/playermark2playername.pickle')
def __init__(self,
pm,
RoP,
csv=False): # 构造参数：球员唯一识别号，常规赛or季后赛('regular' or 'playoff')
# print(pm)
self.pm = pm
self.RoP = 0 if RoP == 'regular' else 1
if not csv:
plyrdr = 'D:/sunyiwu/stat/data/players/%s/%sGames/%sGameBasicStat.pickle' % (
pm, RoP, RoP)
if os.path.exists(plyrdr):
self.dt = regular_items_en if not self.RoP else playoff_items_en
self.ucgd_its = [12, 13] # unchanged_items 前几列数据项雷打不动
self.exists = True
self.plyrFD = plyrdr # 球员个人文件目录
self.data = LoadPickle(self.plyrFD)
if not isinstance(self.data, list):
self.data = self.data.reset_index(drop=True)
if len(self.data.columns) == len(self.dt):
self.data.columns = list(self.dt.keys())
else:
self.data.columns[:self.ucgd_its[self.RoP]] = list(
self.dt.keys())[:self.ucgd_its[self.RoP]]
self.cols = list(self.data.columns)
# print(self.pm, self.cols)
self.seasons = np.sum(self.data['G'] == '1') # 赛季数
self.games = self.data['G'].notna().shape[0]
else:
self.exists = False
def __init__(self):
self.fontsize = 10
self.col_w = 15
self.radiobutton_w = 5
self.paddingx = 10
self.paddingy = 8
self.pm2pn = LoadPickle('../data/playermark2playername.pickle')
self.pn2pm = dict(zip(self.pm2pn.values(), self.pm2pn.keys()))
self.RoP_dict = {'regular': '常规赛', 'playoff': '季后赛'}
self.wd_ = Tk()
self.wd_.title('球员数据查询器')
self.wd_.iconbitmap('../images/nbahalfcourt.ico')
# wd.resizable(width=True, height=True)
self.wd_.geometry('+500+20')
self.stats_setting = {} # 保存查询条件设置
self.plyr_ent_value = StringVar()
self.plyr_ent_value.set('LeBron James')
self.RoP = StringVar()
self.RoP.set(None)
self.AoS = StringVar()
self.AoS.set(None)
self.scope = StringVar()
self.PON = StringVar()
self.PON.set('yes')
self.wd = Frame(self.wd_)
self.wd.grid()
self.wd_.bind("<Return>", self.search_enter)
self.plyr = None
self.plyr_ent = None
def __init__(self):
self.fontsize = 10
self.font = ('SimHei', self.fontsize)
self.col_w = 15
self.radiobutton_w = 5
self.paddingx = 10
self.paddingy = 10
self.pm2pn = LoadPickle('../data/playermark2playername.pickle')
self.pn2pm = dict(zip(self.pm2pn.values(), self.pm2pn.keys()))
self.ROP_dict = {'regular': '常规赛', 'playoff': '季后赛'}
self.wd_ = Tk()
self.wd_.title('比赛数据查询器')
self.wd_.iconbitmap('../images/nbahalfcourt.ico')
# wd.resizable(width=True, height=True)
self.wd_.geometry('+500+20')
self.comboboxs = []
self.stats = []
self.season = StringVar()
self.ROP = StringVar()
self.ROP.set(None)
self.qtr = StringVar()
self.qtr.set('whole')
self.sels = []
self.wd = Frame(self.wd_)
self.wd.grid()
self.wd_.bind("<Return>", self.search_enter)
self.radioframe = None
def __init__(self, startYear, ROP):
self.ROP = 'playoffs' if ROP == 'playoff' else ROP
filename = 'seasonRegularSummary.pickle'\
if ROP == 'regular'\
else 'seasonPlayoffSummary.pickle'
self.gamelist = LoadPickle('./data/seasons/%d_%d/%s'
% (startYear, startYear + 1, filename))
self.gameNos = len(self.gamelist) - 1
def search_by_career(self, stats, ave=1):
c = LoadPickle('../data/players/%s/%sGames/%sSaCAaS.pickle' %
(self.pm, 'playoff' if self.RoP else 'regular',
'playoff' if self.RoP else 'regular'))
resL = list(c.values[-2:][ave - 1])
sentence = self.special_filter(resL, stats)
if sentence and eval(sentence):
return [resL]
else:
return []
def display_pbp(self):
gm = self.gm.get()
if gm:
try:
self.gameflow = LoadPickle(gameMarkToDir(gm, 'regular'))
self.file_name = gameMarkToDir(gm, 'regular')
except:
self.gameflow = LoadPickle(gameMarkToDir(gm, 'playoff'))
self.file_name = gameMarkToDir(gm, 'playoff')
try:
box = LoadPickle(gameMarkToDir(gm, 'regular', tp=2))
except:
box = LoadPickle(gameMarkToDir(gm, 'playoff', tp=2))
# self.columns[1], self.columns[5] = ts[0], ts[1]
print(box[0])
self.qtrs = len(self.gameflow)
self.qtr_btns = []
for i in range(self.qtrs):
self.qtr_btns.append(
Radiobutton(self.qtr_fm,
text=self.qtr_text[i],
value=i,
command=self.qtr_dsp,
variable=self.qtr,
width=10,
height=2))
self.qtr_btns[i].grid(row=1, column=i)
if self.title == 'play-by-play修改器':
self.qtr.set('0')
else:
self.qtr.set(str(int(self.title[1]) - 1))
self.RoW()
self.qtr_dsp(ori=1)
else:
messagebox.showinfo('提示', '请输入game mark！')
pass
def search_by_season(self, stats, ave=1):
c = LoadPickle('../data/players/%s/%sGames/%sSaCAaS.pickle' %
(self.pm, 'playoff' if self.RoP else 'regular',
'playoff' if self.RoP else 'regular'))
tmp, resL, ys = [], [], []
c = list(c.values)
t = 0 if ave else 1
for i, x in enumerate(c[:-2]):
if i % 2 == t:
tmp.append(x)
[ys.append(y) for y in self.yieldSeasons(sgames=False)]
for i, game in enumerate(tmp):
sentence = self.special_filter(game, stats)
if sentence and eval(sentence):
resL.append(['%s %s' % (pm2pn[self.pm], ys[i]), list(game)])
return resL
def teamsWL(ss, ROP):
seasonbsdir = './data/seasons_boxscores/%s/%s/' % (ss, ROP)
gms = [x[:-17] for x in os.listdir(seasonbsdir)]
pc = {} # key: value '队名': [n, n, ..., n] 列表长度为赛季比赛场次，从0开始累加，胜场+1，负场+0
if ROP == 'playoffs':
ot = {} # 记录季后赛对手球队
for gm in gms:
bs = LoadPickle(seasonbsdir + gm + '_boxscores.pickle')
gr = wl(bs[0])
for i, t in enumerate(gr):
if t not in pc:
pc[t] = []
# pc[t].append(pc[t][-1] + -1 if i else pc[t][-1] + 1) if pc[t] else pc[t].append(-1 if i else 1)
pc[t].append(pc[t][-1] if i else pc[t][-1] +
1) if pc[t] else pc[t].append(
0 if i else 1) # 胜队+1，负队+0
if ROP == 'playoffs':
ot = rec_op(gr[0], gr[1], ot)
ot = rec_op(gr[1], gr[0], ot)
if ROP == 'playoffs':
return pc, ot
else:
return pc
def __init__(self,
RoP,
start_season,
end_season,
diffplus=5,
diffminus=-5,
lastmins='5:00.0'):
self.ss = start_season
self.es = end_season
self.dp = diffplus
self.dm = diffminus
self.lm = lastmins
self.RoP = 0 if RoP == 'regular' else 'playoff'
self.pm2pn = LoadPickle('./data/playermark2playername.pickle')
self.columns = [
'score', 'freeThrowPct', 'freeThrowMade', 'freeThrowAttempts',
'twoPtPct', 'twoPtMade', 'twoPtAttempts', 'threePtPct',
'threePtMade', 'threePtAttempts', 'fieldGoalPct', 'fieldGoalMade',
'fieldGoalAttempts', 'twoPtAsted', 'threePtAsted',
'fieldGoalAsted', 'ast 2Pt', 'ast 3Pt', 'eFG%', 'TS%', 'ast score',
'overall score', 'games'
]
self.plyrcgs = {
} # 用于统计球员关键时刻出战场次（暂时无法统计“隐身”情况，即球员在场但无任何行为记录，因此此数字小于等于实际场次）
else:
return pc
def find_no1_wl(wl):
maxp = 0
maxt = ''
for i in wl:
tmp = wl[i][0] / (wl[i][0] + wl[i][1])
if tmp > maxp:
maxp = tmp
maxt = i
return maxp, maxt
chps = LoadPickle('./data/champions.pickle')
op_wl = {}
help_defeat_wl = {}
diff2topwl = {}
for season in tqdm(range(1949, 2020)):
ss = '%d_%d' % (season, season + 1)
cc = chps[ss]
# print(ss)
pc = teamsWL(ss, 'regular')
wls_reg = {}
for tm in pc.keys():
wls_reg[tm] = np.array([pc[tm][-1], len(pc[tm]) - pc[tm][-1]])
no1p, no1t = find_no1_wl(wls_reg)
ccp = wls_reg[cc][0] / (wls_reg[cc][0] + wls_reg[cc][1])
diff2topwl[ss] = [cc, ccp, no1t, no1p, ccp - no1p]
sys.path.append('../')
import os
from tqdm import tqdm
import numpy as np
import pandas as pd
import math
from util import gameMarkToSeason, gameMarkToDir, LoadPickle, writeToPickle, read_nba_pbp
from klasses.Game import Game
from klasses.miscellaneous import MPTime
from windows.result_windows import ShowSingleGame
import matplotlib.pyplot as plt
from scipy.stats.stats import pearsonr
pp = []
wp = LoadPickle('D:/sunyiwu/stat/data/winning_point.pickle')
sss = wp[1]
keys = list(sss)
for key in keys:
sss[key][0] /= sss[key][-1]
sss[key][2] /= sss[key][-1]
# =============================================================================
# items = ['pts off tovs', 'pts% off tovs', '2nd chance pts', '2nd chance pts%',
# 'pts in the paint', 'pts% in the paint', 'biggest lead', 'biggest lead pts%',
# 'longest run', 'longest run pts%', 'eFG%', 'TOV%', 'ORB%', 'FT/FGA',
# '2PT %', '2PT FGA%', '2PT PTS%', '2PT ASTed%',
# '3PT %', '3PT FGA%', '3PT PTS%', '3PT ASTed%',
# 'FT %', 'FTA/FGA', 'FT PTS%']
# chain = {}
class searchByPlyr(object):
def __init__(self):
self.fontsize = 10
self.col_w = 15
self.radiobutton_w = 5
self.paddingx = 10
self.paddingy = 8
self.pm2pn = LoadPickle('../data/playermark2playername.pickle')
self.pn2pm = dict(zip(self.pm2pn.values(), self.pm2pn.keys()))
self.RoP_dict = {'regular': '常规赛', 'playoff': '季后赛'}
self.wd_ = Tk()
self.wd_.title('球员数据查询器')
self.wd_.iconbitmap('../images/nbahalfcourt.ico')
# wd.resizable(width=True, height=True)
self.wd_.geometry('+500+20')
self.stats_setting = {} # 保存查询条件设置
self.plyr_ent_value = StringVar()
self.plyr_ent_value.set('LeBron James')
self.RoP = StringVar()
self.RoP.set(None)
self.AoS = StringVar()
self.AoS.set(None)
self.scope = StringVar()
self.PON = StringVar()
self.PON.set('yes')
self.wd = Frame(self.wd_)
self.wd.grid()
self.wd_.bind("<Return>", self.search_enter)
self.plyr = None
self.plyr_ent = None
def label(self, win, text, fs, w, h, ac):
return Label(win,
text=text,
font=('SimHei', fs),
width=w,
height=h,
anchor=ac)
def one_tick_sel(self, win, text, value, w, variable, command):
return Radiobutton(win,
text=text,
value=value,
command=command,
variable=variable,
width=w,
height=1)
def place_sep(self, win, row, columnspan=12): # 分割线
sep = ttk.Separator(win, orient='horizontal')
sep.grid(padx=3,
pady=3,
row=row,
column=0,
columnspan=columnspan,
sticky='ew')
def place_equal_inf(self, row): # 信息数据Frame
equal_fm = Frame(self.wd)
equal_fm.grid(row=row, column=0, columnspan=12)
ls = ['Tm', 'Opp', 'GS', 'RoH', 'WoL'] if self.RoP.get() == 'regular'\
else ['Tm', 'Opp', 'GS', 'RoH', 'WoL', 'Series']
self.place_stat(equal_fm, ls, 'i // 3 + 1', 'i % 3 * 2', type=0)
def place_basic(self, row): # 基础数据Frame
basic_fm = Frame(self.wd)
basic_fm.grid(row=row, column=0, columnspan=12)
ls = [
'PTS', 'TRB', 'AST', 'BLK', 'STL', 'TOV', 'ORB', 'DRB', 'PF', 'MP',
'GmSc', '+/-'
]
self.place_stat(basic_fm, ls, 'i // 3 + 1', 'i % 3 * 4')
def place_shooting(self, row): # 投篮数据Frame
shooting_fm = Frame(self.wd)
shooting_fm.grid(row=row, column=0, columnspan=12)
ls = ['FG', 'FGA', 'FG%', '3P', '3PA', '3P%', 'FT', 'FTA', 'FT%']
self.place_stat(shooting_fm, ls, 'i // 3 + 1', 'i % 3 * 4')
def place_inf(self, row): # 信息比较数据Frame
inf_fm = Frame(self.wd)
inf_fm.grid(row=row, column=0, columnspan=12)
ls = ['G', 'Date', 'Age', 'Diff'] if self.RoP.get() == 'regular'\
else ['G', 'Playoffs', 'G#', 'Diff']
self.place_stat(inf_fm, ls, 'i // 2 + 1', 'i % 2 * 4')
def place_stat(self, win, ls, row_, col_, type=1): # type 1:大于小于 0:等于
# ==============查询条件输入框布局====================
for i, k in enumerate(ls):
row, col = eval(row_), eval(col_)
if type:
self.label(win, en2ch[k][0] + ':', self.fontsize, self.col_w,
1, 'e').grid(padx=self.paddingx,
pady=self.paddingy,
row=row,
column=col)
ent1, ent2 = Entry(win, width=10), Entry(win, width=10)
self.label(win, '-', self.fontsize, self.col_w // 3, 1,
'center').grid(padx=self.paddingx // 2,
pady=self.paddingy,
row=row,
column=col + 2)
ent1.grid(row=row,
column=col + 1,
padx=self.paddingx // 2,
pady=self.paddingy)
ent2.grid(row=row,
column=col + 3,
padx=self.paddingx // 2,
pady=self.paddingy)
self.stats_setting[k] = [ent1, ent2]
else:
self.label(win, en2ch[k][0] + ':', self.fontsize, 12, 1,
'center').grid(padx=self.paddingx,
pady=self.paddingy,
row=row,
column=col)
ent = Entry(win, width=10)
ent.grid(row=row,
column=col + 1,
padx=self.paddingx,
pady=self.paddingy)
self.stats_setting[k] = eval('[ent1, ent2]') if type else eval(
'[ent]')
def RoPselection(self):
[i.grid_remove() for i in self.wd.grid_slaves()[:-10]]
self.place_equal_inf(3)
self.place_sep(self.wd, 4)
self.place_basic(5)
self.place_sep(self.wd, 6)
self.place_shooting(7)
self.place_sep(self.wd, 8)
self.place_inf(9)
def AoSselection(self):
pass
def plyr_or_not(self): # 是否按球员分组
self.plyr['text'] = '最小场数' if self.PON.get() == 'no' else '球员'
self.plyr_ent_value.set('1') if self.PON.get(
) == 'no' else self.plyr_ent_value.set('LeBron James')
def search_enter(self, event): # 绑定回车键触发搜索函数
self.search()
def search(self): # 点击按钮触发搜索函数
if self.stats_setting:
stats = {}
for k in self.stats_setting.keys(): # 遍历文本框控件，收集查询条件
tp = 1 if en2ch[k][1] == 1 or k == 'WoL' or k == 'MP' else 0
ent_s = self.stats_setting[k]
if len(ent_s) == 1: # 相等比较
if ent_s[0].get():
stats[k] = [
' == "%s" and ' %
ent_s[0].get() if tp else ' == %s and ' %
ent_s[0].get()
]
else:
if ent_s[0].get() and ent_s[1].get(): # 大于小于同时存在
stats[k] = [
' >= "%s" and ' %
ent_s[0].get() if tp else ' >= %s and ' %
ent_s[0].get(),
' <= "%s" and ' %
ent_s[1].get() if tp else ' <= %s and ' %
ent_s[1].get()
]
elif ent_s[0].get() and not ent_s[1].get(): # 只有大于
stats[k] = [
' >= "%s" and ' %
ent_s[0].get() if tp else ' >= %s and ' %
ent_s[0].get()
]
elif not ent_s[0].get() and ent_s[1].get(): # 只有小于
stats[k] = [
' <= "%s" and ' %
ent_s[1].get() if tp else ' <= %s and ' %
ent_s[1].get()
]
if not stats:
messagebox.showinfo('提示', '请设置查询条件！')
else:
if self.PON.get() == 'yes': # 单球员查询
if self.plyr_ent_value.get() not in self.pn2pm.keys():
messagebox.showinfo('提示', '球员姓名不存在！')
return
player = Player(self.pn2pm[self.plyr_ent_value.get()],
self.RoP.get())
if self.scope.get() == '单场比赛':
pt = 1
res = player.search_by_game(stats)
elif self.scope.get() == '赛季':
pt = 2
res = player.search_by_season(stats,
int(self.AoS.get()))
elif self.scope.get() == '职业生涯':
pt = 2
res = player.search_by_career(stats,
int(self.AoS.get()))
res = [[self.plyr_ent_value.get(), x] for x in res]
elif self.scope.get() == '连续比赛':
pt = 0
res = player.search_by_consecutive(stats)
else: # 按球员分组查询
pt = 0
# start = time.time()
# pp = [[x, self.RoP.get(), stats] for x in list(self.pm2pn.keys())]
# print(time.time() - start)
# pool = ThreadPool(4)
# tmp = pool.map(process, pp)
# pool.close()
# pool.join()
# print(time.time() - start)
# res = []
# for x in tmp:
# if x:
# res.append(x)
# print(time.time() - start)
start = time.time()
res = []
min_ = 50 if self.RoP.get() == 'regular' else 5
for p in tqdm(list(self.pm2pn.keys())):
player = Player(p, self.RoP.get())
if player.exists and not isinstance(
player.data, list) and player.games > min_:
if self.scope.get() == '单场比赛':
tmp = player.search_by_game(
stats, minG=int(self.plyr_ent_value.get()))
if tmp:
res.append([self.pm2pn[p], tmp])
elif self.scope.get() == '赛季':
pt = 2
tmp = player.search_by_season(
stats, int(self.AoS.get()))
res += tmp
elif self.scope.get() == '职业生涯':
pt = 2
tmp = player.search_by_career(
stats, int(self.AoS.get()))
if tmp:
tmp = [[self.pm2pn[p], x] for x in tmp]
res += tmp
elif self.scope.get() == '连续比赛':
pt = 0
res += player.search_by_consecutive(
stats, minG=int(self.plyr_ent_value.get()))
print(time.time() - start)
# 处理结果
if res:
RP = regular_items_en if self.RoP.get(
) == 'regular' else playoff_items_en
win_klass = ShowSingleResults if pt else ShowGroupResults
print(win_klass)
if pt == 2:
win_klass = ShowResults
result_window = win_klass(res,
list(RP.keys()),
self.RoP.get(),
detail=False if pt else True,
AoS=int(self.AoS.get()))
win_title = '%s %s 查询结果（按%s）' % (self.plyr_ent_value.get(),
self.RoP_dict[self.RoP.get()], self.scope.get())\
if pt else '%s查询结果（按%s）' % ('常规赛' if self.RoP.get() == 'regular' else '季后赛', self.scope.get())
result_window.title(win_title)
num = (len(res) - 2 * pt) if pt != 2 else len(res)
result_window.loop('共查询到%d组数据' % num, stats)
else:
messagebox.showinfo('提示', '未查询到符合条件的数据！')
else:
messagebox.showinfo('提示', '请选择比赛类型！')
def loop(self):
# ==============背景图片====================
bg_img = Image.open('../images/james.jpg')
bg_img = bg_img.resize(
(int(bg_img.size[0] * 0.95), int(bg_img.size[1] * 0.95)))
bg_img = cv2.cvtColor(np.asarray(bg_img), cv2.COLOR_RGB2BGR)
bg_img = cv2.copyMakeBorder(bg_img, 0, 0, 100, 100,
cv2.BORDER_REFLECT_101)
bg_img = Image.fromarray(cv2.cvtColor(bg_img, cv2.COLOR_BGR2RGB))
bg_img.putalpha(32) # 透明度
bg_img = ImageTk.PhotoImage(bg_img)
Label(self.wd, image=bg_img).place(x=0, y=0, relwidth=1, relheight=1)
# ==============球员名称label&entry====================
self.plyr = self.label(self.wd, '球员:', self.fontsize, self.col_w, 1,
'e')
self.plyr_ent = Entry(self.wd,
width=20,
textvariable=self.plyr_ent_value,
font=('SimHei', 15))
self.plyr.grid(padx=self.paddingx,
pady=self.paddingy,
row=1,
rowspan=2,
column=0)
self.plyr_ent.grid(padx=self.paddingx,
pady=self.paddingy,
row=1,
rowspan=2,
column=1,
columnspan=3)
# ==============单球员or多球员单选Frame====================
plyr_fm = Frame(self.wd, height=1)
plyr_yes = self.one_tick_sel(plyr_fm, '单球员查询', 'yes',
self.radiobutton_w * 2, self.PON,
self.plyr_or_not)
plyr_no = self.one_tick_sel(plyr_fm, '按球员分组', 'no',
self.radiobutton_w * 2, self.PON,
self.plyr_or_not)
plyr_fm.grid(padx=self.paddingx,
pady=self.paddingy,
row=1,
rowspan=2,
column=4,
columnspan=3)
plyr_yes.grid(row=1, column=0)
plyr_no.grid(row=1, column=1)
# ==============比赛类型组合选择Frame====================
gametype_fm = Frame(self.wd)
rglr_sel = self.one_tick_sel(gametype_fm, '常规赛', 'regular',
self.radiobutton_w, self.RoP,
self.RoPselection)
plyf_sel = self.one_tick_sel(gametype_fm, '季后赛', 'playoff',
self.radiobutton_w, self.RoP,
self.RoPselection)
scope_sel = ttk.Combobox(gametype_fm, width=7, textvariable=self.scope)
scope_sel['value'] = ['单场比赛', '赛季', '职业生涯', '连续比赛', '连续赛季']
scope_sel.current(0)
ave_sel = self.one_tick_sel(gametype_fm, '场均', '1', 2, self.AoS,
self.AoSselection)
sum_sel = self.one_tick_sel(gametype_fm, '总和', '0', 2, self.AoS,
self.AoSselection)
self.AoS.set('1')
self.place_sep(gametype_fm, 2, columnspan=2)
gametype_fm.grid(padx=self.paddingx,
pady=self.paddingy,
row=1,
rowspan=2,
column=7,
columnspan=4)
rglr_sel.grid(padx=self.paddingx, pady=self.paddingy, row=1, column=0)
plyf_sel.grid(padx=self.paddingx, pady=self.paddingy, row=1, column=1)
scope_sel.grid(padx=self.paddingx,
pady=self.paddingy,
row=1,
rowspan=3,
column=2,
columnspan=2)
ave_sel.grid(padx=self.paddingx,
pady=self.paddingy,
row=3,
column=0,
sticky='e')
sum_sel.grid(padx=self.paddingx,
pady=self.paddingy,
row=3,
column=1,
sticky='w')
# ==============搜索button====================
search_img = Image.open('../images/kobe_dunk.jpg')
search_img = ImageTk.PhotoImage(search_img)
search_button = Button(self.wd,
text='查 询',
width=60,
height=30,
image=search_img,
compound='center',
cursor='hand2',
command=self.search,
font=('SimHei', 14))
search_button.grid(padx=self.paddingx,
pady=self.paddingy,
row=1,
rowspan=2,
column=11)
# ==============说明label====================
self.place_sep(self.wd, 20)
notion = self.label(self.wd, '说明:', self.fontsize, self.col_w, 1, 'e')
help_note = '主客场：0客1主 比赛序号：赛季第n场比赛\n' \
'年龄：35-001 赛果：W/L 日期：%s\n' \
'季后赛轮次：E/WC1->第一轮，E/WCS->分区半决赛，' \
'E/WCF->分区决赛，FIN->总决赛' % time.strftime("%Y%m%d")
notion_ = Label(self.wd,
text=help_note,
font=('SimHei', 11),
width=self.col_w * 5,
height=6,
anchor='w',
justify='left')
notion.grid(padx=self.paddingx, pady=self.paddingy, row=26, column=0)
notion_.grid(padx=self.paddingx,
pady=self.paddingy,
row=26,
rowspan=6,
column=1,
columnspan=11,
sticky='w')
# wd.attributes("-alpha", 0.8)
self.RoP.set('regular')
self.RoPselection()
self.wd.mainloop()
#%%
regularOrPlayoff = 1
ROP = regularOrPlayoffs[regularOrPlayoff]
for i in players:
print("Analysing player: %s ..." % i[0])
player = Player(i[1], ROP)
missing_shot = 0
areaPct = np.zeros((21,3)) #0miss1make
for season in player.yieldSeasons():
for gameInf in player.yieldGames(season):
gameMark, team = gameInf[1], gameInf[3]
# 读取比赛文件
gameDir = gameMarkToDir(gameMark, ROP)
gameDir_shot = gameMarkToDir(gameMark, ROP, shot=True)
playerGame = LoadPickle(gameDir)
playerGame_shot = LoadPickle(gameDir_shot)
index_shot = 0
# 判断主客场
HOA = 1 if team == gameMark[-3:] else 0 # 0客1主
HomeOrAt = HomeOrAts[HOA-1]
for index, quarter in enumerate(playerGame):
for play in quarter:
if len(play) == 6:
if play[HomeOrAt[1]]:
playerPlayed = play[HomeOrAt[1]].split(' ')[0]
if playerPlayed == i[1] and (play[HomeOrAt[0]] or 'misses' in play[HomeOrAt[1]]):
# 此球员有得分/错失得分
score = 0
diff = play[3]
if HOA == 1:
def __init__(self, gm, ROP, HOA=None):
self.gamemark = gm
self.gameflow = LoadPickle(gameMarkToDir(gm, ROP, shot=True))
self.HOA = HOA if HOA != None else None
class Player(object):
def __init__(self,
pm,
RoP,
csv=False): # 构造参数：球员唯一识别号，常规赛or季后赛('regular' or 'playoff')
# print(pm)
self.pm = pm
self.RoP = 0 if RoP == 'regular' else 1
if not csv:
plyrdr = 'D:/sunyiwu/stat/data/players/%s/%sGames/%sGameBasicStat.pickle' % (
pm, RoP, RoP)
if os.path.exists(plyrdr):
self.dt = regular_items_en if not self.RoP else playoff_items_en
self.ucgd_its = [12, 13] # unchanged_items 前几列数据项雷打不动
self.exists = True
self.plyrFD = plyrdr # 球员个人文件目录
self.data = LoadPickle(self.plyrFD)
if not isinstance(self.data, list):
self.data = self.data.reset_index(drop=True)
if len(self.data.columns) == len(self.dt):
self.data.columns = list(self.dt.keys())
else:
self.data.columns[:self.ucgd_its[self.RoP]] = list(
self.dt.keys())[:self.ucgd_its[self.RoP]]
self.cols = list(self.data.columns)
# print(self.pm, self.cols)
self.seasons = np.sum(self.data['G'] == '1') # 赛季数
self.games = self.data['G'].notna().shape[0]
else:
self.exists = False
def season_index(self, games):
# print(type(games['G'][0]), games['G'])
return list(games[games['G'] == '1'].index) + [games.shape[0]]
def yieldSeasons(self, sgames=True): # 按赛季返回 sgames为False时只返回赛季年号
games = self.on_board_games(self.data)
games = games.reset_index(drop=True)
ss_ix = self.season_index(games)
for i in range(self.seasons):
ss = games.iloc[ss_ix[i]]['Playoffs' if self.RoP else 'Date']
yy = ss[:4]
yy = '%s_%s' % (str(int(yy) - 1), yy) if self.RoP else '%s_%s' % (
yy, str(int(yy) + 1))
if sgames:
yield games[ss_ix[i]:ss_ix[i + 1]], yy
else:
yield yy
def getSeason(self, ss):
gmcol = 'Playoffs' if self.RoP else 'Date'
start = ss[:4]
gms = self.data[self.data['G'].notna()]
s, e, flag = -1, -1, 0
for i in gms.index:
if not flag and gms[gmcol][i] > start + '10':
s = i
flag = 1
elif s != -1 and gms['G'][i] == '1':
e = i - 1
break
res = gms.loc[s:e] if e != -1 else gms.loc[s:]
res = res.reset_index(drop=True)
return res
def on_board_games(self, games):
# games = games['G'].astype('str')
# games = games[games['G'] != '']
return games[games['G'].notna()]
def yieldGames(self, season, del_absent=True): # 按单场比赛返回
if del_absent:
season = self.on_board_games(season)
for i in season.index:
yield season.loc[i]
def _get_item(self, item, season_index=None):
games = self.data[self.ss_ix[season_index - 1]:self.
ss_ix[season_index]] if season_index else self.data
return self.on_board_games(games)[item]
def seasonAVE(self, ind, item):
# 求取赛季平均，传入参数：赛季序号、统计项名称
return np.mean(
self._get_item(item, season_index=ind).astype(np.float64))
def average(self, item):
return np.mean(self._get_item(item).astype(np.float64))
def seasonSUM(self, ind, item):
# 求取赛季平均，传入参数：赛季序号、统计项名称
return np.sum(
self._get_item(item, season_index=ind).astype(np.float64))
def sum(self, item):
return np.sum(self._get_item(item).astype(np.float64))
def special_filter(self, game, stats): # 按条件筛选每条数据
sentence = ''
for k in stats:
if en2ch[k][1] == 0: # 数值型
x = float(game[self.dt[k]])
if math.isnan(x):
return ''
elif en2ch[k][1] == 1: # 字符型
x = '"%s"' % str(game[self.dt[k]])
else: # 特殊处理
if k == 'WoL':
x = '"%s"' % game[7 if self.RoP else 6][0]
elif k == 'RoH':
x = '0' if game[4] == '@' else '1'
elif k == 'Date' or k == 'Playoffs':
x = '%s' % game[1][:8]
elif k == 'MP':
tmp = game[9 if self.RoP else 8]
if not isinstance(tmp, str):
continue
if int(tmp[:tmp.index(':')]) < 10:
tmp = '0' + tmp # 分钟数小于10前面加0
if len(tmp) == 5: # 没有数秒位
tmp += ':00'
x = '"%s"' % tmp
else:
x = '%d' % int(game[7 if self.RoP else 6][3:-1])
for cmp_i, t in enumerate(stats[k]):
sentence += '%s%s' % (x, stats[k][cmp_i])
sentence = sentence[:-5]
return sentence
def ave_and_sum(self, tmp, type=2): # type=0:计算总和 1:计算均值 2:计算均值和总和
if isinstance(tmp, list):
tmp = pd.DataFrame(tmp,
columns=regular_items_en.keys()
if not self.RoP else playoff_items_en.keys())
if type:
ave = []
sumn = []
for k in tmp.columns:
if k == 'G': # 首列
if type:
ave.append('%d场平均' % tmp.shape[0])
sumn.append('%d场总和' % tmp.shape[0])
elif k in ['Playoffs', 'Date', 'Age', 'Tm', 'Opp', 'Series', 'G#']:
if type:
ave.append('/')
sumn.append('/')
elif k == 'RoH': # 统计主客场数量
try:
at = np.sum(tmp[k] == '@')
except:
at = 0
if type:
ave.append('%dR/%dH' % (at, tmp.shape[0] - at))
sumn.append('%dR/%dH' % (at, tmp.shape[0] - at))
elif k == 'WoL': # 统计几胜几负
origin = WinLoseCounter(False)
for i in tmp[k]:
origin += WinLoseCounter(True, strwl=i)
if type:
ave.append(origin.average())
sumn.append(origin)
elif k == 'GS': # 统计首发次数
cs = tmp[k].value_counts()
if cs.empty:
if type:
ave.append('')
sumn.append('')
else:
try:
s = np.sum(tmp[k] == '1')
except:
s = 0
if type:
ave.append('%d/%d' % (s, tmp.shape[0]))
sumn.append('%d/%d' % (s, tmp.shape[0]))
elif k == 'MP': # 时间加和与平均
sum_time = MPTime('0:00.0', reverse=False)
for i in tmp[k]:
if isinstance(i, str):
sum_time += MPTime(i, reverse=False)
if type:
ave.append(sum_time.average(tmp.shape[0]))
sumn.append(sum_time.strtime[:-2])
elif '%' in k: # 命中率单独计算
goal = tmp[[k[:-1], k[:-1] + 'A']]
goal = goal.dropna(axis=0, how='any')
goal = goal.astype('float').sum(axis=0)
p = goal[k[:-1]] / goal[k[:-1] +
'A'] if goal[k[:-1] +
'A'] else float('nan')
if not math.isnan(p):
p = '%.3f' % p
if p != '1.000':
p = p[1:]
else:
p = ''
if type:
ave.append(p)
sumn.append(p)
else:
tmp_sg = tmp[k][tmp[k].notna()]
# tmp_sg = tmp_sg[tmp_sg != '']
if not tmp_sg.empty:
if type:
# print(tmp_sg)
try:
a = '%.1f' % tmp_sg.astype('float').mean()
except:
tmp_sg = tmp_sg[tmp_sg != '']
a = '%.1f' % tmp_sg.astype('float').mean()
# 除比赛评分以外其他求和结果均为整数
s = '%.1f' % tmp_sg.astype('float').sum(
) if k == 'GmSc' else int(tmp_sg.astype('int').sum())
if k == '+/-': # 正负值加+号
if type and s != 0 and a[0] != '-':
a = '+' + a
if s > 0:
s = '+%d' % s
if s == 'nan':
if type:
a = ''
s = ''
else:
a, s = '', ''
if type:
ave.append(a)
sumn.append(s)
if type == 0:
return [sumn]
elif type == 1:
return [ave]
else:
return [ave, sumn]
def search_by_consecutive(self, stats, minG=2):
if minG < 2:
minG = 2
# print(stats)
resL, tmp = [], []
for ss, ys in self.yieldSeasons():
for game in list(ss.values):
sentence = self.special_filter(game, stats)
if sentence and eval(sentence):
tmp.append(game)
else:
if len(tmp) >= minG:
tmp += self.ave_and_sum(tmp)
resL.append(['%s %s' % (pm2pn[self.pm], ys), tmp])
tmp = []
return resL
def search_by_season(self, stats, ave=1):
c = LoadPickle('../data/players/%s/%sGames/%sSaCAaS.pickle' %
(self.pm, 'playoff' if self.RoP else 'regular',
'playoff' if self.RoP else 'regular'))
tmp, resL, ys = [], [], []
c = list(c.values)
t = 0 if ave else 1
for i, x in enumerate(c[:-2]):
if i % 2 == t:
tmp.append(x)
[ys.append(y) for y in self.yieldSeasons(sgames=False)]
for i, game in enumerate(tmp):
sentence = self.special_filter(game, stats)
if sentence and eval(sentence):
resL.append(['%s %s' % (pm2pn[self.pm], ys[i]), list(game)])
return resL
def search_by_career(self, stats, ave=1):
c = LoadPickle('../data/players/%s/%sGames/%sSaCAaS.pickle' %
(self.pm, 'playoff' if self.RoP else 'regular',
'playoff' if self.RoP else 'regular'))
resL = list(c.values[-2:][ave - 1])
sentence = self.special_filter(resL, stats)
if sentence and eval(sentence):
return [resL]
else:
return []
# def process(self, game, stats):
# sentence = self.special_filter(game, stats)
# if sentence and eval(sentence):
# return game
def search_by_game(self, stats, minG=1):
if minG < 1:
minG = 1
# 可统一比较大小或判断相等
games = self.on_board_games(self.data).values
if len(games) >= minG:
resL = []
# pool = Pool(processes=4)
# multiprocessing.set_start_method('spawn')
# for game in games:
# res = pool.apply(self.process, (game, stats, ))
# if res is not None:
# resL.append(res)
# pool.close()
# pool.join()
for game in games:
sentence = self.special_filter(game, stats)
if sentence and eval(sentence):
resL.append(game)
if len(resL) < minG:
return []
else:
resL += self.ave_and_sum(resL) # 求取平均和总和
return resL
else:
return []
# -*- coding:utf8 -*-
import sys
sys.path.append('../')
from util import LoadPickle, playerMarkToDir
from klasses.miscellaneous import MPTime
from klasses.Player import Player
from klasses.Game import Game
import pandas as pd
import numpy as np
from tqdm import tqdm
pd.options.display.expand_frame_repr = False
pd.options.display.width = 50
RoF = ['regular', 'playoff']
pm2pn = LoadPickle('./data/playermark2playername.pickle')
plyr_ss = LoadPickle('./data/playerSeasonAverage.pickle') # 球员所在球队及对手球队赛季场均数据
lg_ss = LoadPickle('./data/leagueSeasonAverage.pickle') # 联盟球队赛季场均数据
games_td = [30, 5]
tartext = ['Steal', 'Turnover', 'Block', 'Assist', 'OffReb']
enforce = {} # 球员生涯/赛季数据影响力因子均值（抢断失误盖帽助攻前板）
lg_season = {} # 联盟生涯/赛季数据影响力因子均值（抢断失误盖帽助攻前板）
for tar_item in range(len(tartext)): # 0抢断1失误2盖帽3助攻4前板
enforce[tartext[tar_item]] = LoadPickle(
'./data/Enforce/player%sEnforce.pickle' % tartext[tar_item])
lg_season[tartext[tar_item]] = LoadPickle(
'./data/Enforce/season%sEnforceRecord.pickle' % tartext[tar_item])
# for tar_item in range(len(tartext)):
# print(tartext[tar_item], len(enforce[tartext[tar_item]]))
pms = list(enforce['Turnover'].keys())
#%%
# 0输1赢2总；0全部1投中2投失3全部两分4投中5投失6全部三分7投中8投失
numGames = []
cnt = np.zeros((24, 4, 3, 9))
dis = np.zeros((24, 4, 3, 9))
score = np.zeros((24, 2))
for n, season in enumerate(range(1996, 2020)):
gameDir = './data/seasons_shot/%d_%d/%s/' % (
season, season + 1, regularOrPlayoffs[regularOrPlayoff])
upROP = regularOrPlayoffs[regularOrPlayoff][0].upper(
) + regularOrPlayoffs[regularOrPlayoff][1:]
summaryDir = './data/seasons/%d_%d/season%sSummary.pickle' % (
season, season + 1, upROP)
summary = LoadPickle(summaryDir)
numGames.append(len(summary) - 1)
print('starting to analysing season %s-%s:' % (season, season + 1))
for game in tqdm(summary[1:]):
f = open(gameDir + game[0] + '_shot.pickle', 'rb')
c = pickle.load(f)
f.close()
if season < 2000:
win = 0 if int(game[2]) > int(game[4]) else 1
score[n][0] += min(int(game[2]), int(game[4]))
score[n][1] += max(int(game[2]), int(game[4]))
else:
win = 0 if int(game[3]) > int(game[5]) else 1
score[n][0] += min(int(game[3]), int(game[5]))
score[n][1] += max(int(game[3]), int(game[5]))
class Search_by_game(object):
def __init__(self):
self.fontsize = 10
self.font = ('SimHei', self.fontsize)
self.col_w = 15
self.radiobutton_w = 5
self.paddingx = 10
self.paddingy = 10
self.pm2pn = LoadPickle('../data/playermark2playername.pickle')
self.pn2pm = dict(zip(self.pm2pn.values(), self.pm2pn.keys()))
self.ROP_dict = {'regular': '常规赛', 'playoff': '季后赛'}
self.wd_ = Tk()
self.wd_.title('比赛数据查询器')
self.wd_.iconbitmap('../images/nbahalfcourt.ico')
# wd.resizable(width=True, height=True)
self.wd_.geometry('+500+20')
self.comboboxs = []
self.stats = []
self.season = StringVar()
self.ROP = StringVar()
self.ROP.set(None)
self.qtr = StringVar()
self.qtr.set('whole')
self.sels = []
self.wd = Frame(self.wd_)
self.wd.grid()
self.wd_.bind("<Return>", self.search_enter)
self.radioframe = None
def one_tick_sel(self, text, value, command, win):
return Radiobutton(win, text=text, value=value, command=command,
variable=self.qtr, width=self.radiobutton_w, height=1)
def season_cbox_sel(self, event):
[i.grid_remove() for i in self.radioframe.grid_slaves()]
if self.season.get()[:4] >= '1996':
end = 8
elif self.season.get()[:4] < '1983':
end = 1
else:
end = 2
for ix, i in enumerate(self.sels[:end]):
i.grid(padx=self.paddingx, pady=self.paddingy, row=1, column=ix)
def search_enter(self, event): # 绑定回车键触发搜索函数
self.search()
def search(self):
print('功能尚未开放')
def timespan(self):
pass
def ROPselection(self):
pass
def QTRselection(self):
pass
def loop(self):
# 背景图片
bg_img = Image.open('../images/james2.jpg')
bg_img.putalpha(64)
bg_img = ImageTk.PhotoImage(bg_img)
Label(self.wd, image=bg_img).place(x=0, y=0, relwidth=1, relheight=1)
self.radioframe = Frame(self.wd)
self.radioframe.grid(padx=self.paddingx, pady=self.paddingy, row=4, column=0, columnspan=5, sticky='ew')
# 控件设置
season_label = Label(self.wd, text='赛季:', font=self.font, width=self.col_w, height=1, anchor='e')
season_sel = ttk.Combobox(self.wd, width=9, textvariable=self.season)
season_sel['value'] = ['%d-%d' % (x, x + 1) for x in range(1949, 2020)]
season_sel.current(len(season_sel['value']) - 1)
timespan_frame = LabelFrame(self.wd, text='时间跨度', labelanchor="nw")
sglssn_sel = self.one_tick_sel('单赛季', 'single_season', self.timespan, timespan_frame)
tilnow_sel = self.one_tick_sel('至今', 'tillnow', self.timespan, timespan_frame)
rf_frame = LabelFrame(self.wd, text='比赛类型', labelanchor="nw")
rglr_sel = self.one_tick_sel('常规赛', 'regular', self.ROPselection, rf_frame)
plyf_sel = self.one_tick_sel('季后赛', 'playoff', self.ROPselection, rf_frame)
search_img = Image.open('../images/kobe_dunk.jpg')
search_img = ImageTk.PhotoImage(search_img)
search_button = Button(self.wd, text='查 询', width=65, height=30, image=search_img,
compound='center', cursor='hand2', command=self.search, font=('SimHei', 14))
team_label = Label(self.wd, text='球队:', font=self.font, width=self.col_w, height=1, anchor='e')
oppteam_label = Label(self.wd, text='对手球队:', font=self.font, width=self.col_w, height=1, anchor='e')
team_ent = Entry(self.wd, width=10)
oppteam_ent = Entry(self.wd, width=10)
sel_texts = [['全场', 'whole'], ['进阶', 'advcd'], ['第一节', '1st'], ['第二节', '2nd'],
['上半场', '1hf'], ['第三节', '3rd'], ['第四节', '4th'], ['下半场', '2hf']]
for ts in sel_texts:
self.sels.append(self.one_tick_sel(ts[0], ts[1], self.QTRselection, self.radioframe))
# 控件布局
season_label.grid(padx=self.paddingx, pady=self.paddingy, row=1, rowspan=2, column=0)
season_sel.grid(padx=self.paddingx, pady=self.paddingy, row=1, rowspan=2, column=1)
season_sel.bind("<<ComboboxSelected>>", self.season_cbox_sel)
timespan_frame.grid(padx=self.paddingx, pady=self.paddingy, row=1, rowspan=2, column=2)
sglssn_sel.grid(padx=self.paddingx, pady=self.paddingy, row=1, column=0)
tilnow_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=0)
rf_frame.grid(padx=self.paddingx, pady=self.paddingy, row=1, rowspan=2, column=3)
rglr_sel.grid(padx=self.paddingx, pady=self.paddingy, row=1, column=0)
plyf_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=0)
search_button.grid(padx=self.paddingx, pady=self.paddingy, row=1, rowspan=2, column=4, columnspan=2)
team_label.grid(padx=self.paddingx, pady=self.paddingy, row=3, column=0)
oppteam_label.grid(padx=self.paddingx, pady=self.paddingy, row=3, column=2)
team_ent.grid(padx=self.paddingx, pady=self.paddingy, row=3, column=1)
oppteam_ent.grid(padx=self.paddingx, pady=self.paddingy, row=3, column=3)
# whole_sel.grid(padx=self.paddingx, pady=self.paddingy, row=1, column=2, sticky='ew')
# advcd_sel.grid(padx=self.paddingx, pady=self.paddingy, row=1, column=3, sticky='ew')
# first_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=0, sticky='ew')
# secnd_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=1, sticky='ew')
# fsthf_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=2, sticky='ew')
# third_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=3, sticky='ew')
# forth_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=4, sticky='ew')
# sndhf_sel.grid(padx=self.paddingx, pady=self.paddingy, row=2, column=5, sticky='ew')
self.wd.mainloop()
def seasonfile(self, sn):
file = './data/seasons_RAT/%s.pickle' % sn
return LoadPickle(file)
#!/usr/bin/python
# -*- coding:utf8 -*-
from util import LoadPickle, getCode, writeToPickle
import os
RoFs = ['regular', 'playoffs']
for season in range(2014, 2020):
ss = '%d_%d' % (season, season + 1)
print(ss)
for i in range(2):
season_dir = './data/seasons/%s/%s/' % (ss, RoFs[i])
gms = os.listdir(season_dir)
for gm in gms:
c = LoadPickle(season_dir + gm)
for q, qtr in enumerate(c):
for ix, r in enumerate(qtr):
if len(r) == 6 and 'enters' in (r[1] if r[1] else r[-1]):
ind = 1 if r[1] else 5
tmp = r[ind].split(' ')
pm1, pm2 = tmp[0], tmp[-1]
if pm1 == pm2:
print(gm, c[q][ix][ind])
url = 'https://www.basketball-reference.com/boxscores/pbp/%s.html' % gm[:-7]
plays = getCode(url, 'UTF-8')
plays = plays.find('table', class_='stats_table').find_all('tr')
for play in plays:
tdPlays = play.find_all('td')
if len(tdPlays) == 6:
for p in tdPlays:
if p.find_all('a'):
import sys
sys.path.append('../')
import pandas as pd
import os
from klasses.stats_items import *
from klasses.miscellaneous import MPTime, WinLoseCounter
from util import LoadPickle
import numpy as np
import math
import multiprocessing
from multiprocessing.pool import Pool
pm2pn = LoadPickle('D:/sunyiwu/stat/data/playermark2playername.pickle')
class Player(object):
def __init__(self,
pm,
RoP,
csv=False): # 构造参数：球员唯一识别号，常规赛or季后赛('regular' or 'playoff')
# print(pm)
self.pm = pm
self.RoP = 0 if RoP == 'regular' else 1
if not csv:
plyrdr = 'D:/sunyiwu/stat/data/players/%s/%sGames/%sGameBasicStat.pickle' % (
pm, RoP, RoP)
if os.path.exists(plyrdr):
self.dt = regular_items_en if not self.RoP else playoff_items_en
self.ucgd_its = [12, 13] # unchanged_items 前几列数据项雷打不动
self.exists = True
[[MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0')],
[MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0')],
[MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0'), MPTime('0:00.0')]]]
average_time = [[['', '', '', '', '', '', '', '', ''], ['', '', '', '', '', '', '', '', ''], ['', '', '', '', '', '', '', '', '']],
[['', '', '', '', '', '', '', '', ''], ['', '', '', '', '', '', '', '', ''], ['', '', '', '', '', '', '', '', '']]]
count_score = np.zeros((2, 3, 9))
ss = '%d_%d' % (season, season + 1)
# print(ss)
for i in range(2): # 分别统计常规赛、季后赛
gms = os.listdir('D:/sunyiwu/stat/data/seasons/%s/%s/' % (ss, regularOrPlayoffs[i]))
for gm in tqdm(gms):
count_games[i] += 1
# print('\t\t\t' + gm)
game = Game(gm[:-7], regularOrPlayoffs[i])
gameplyrs = game.teamplyrs()
record = LoadPickle(gameMarkToDir(gm[:-7], regularOrPlayoffs[i], tp=3))
# _, _, _, record = game.game_scanner(gm[:-7])
zoom = 0
bp = -1
pm = ''
for rec in record:
if zoom: # 处于抢断/失误观察期内
# 出现得失分
if 'MK' in rec or 'MS' in rec:
GoM = 1 if 'MK' in rec else 0
KoS = 'MK' if GoM else 'MS'
plyr_side = 0 if pm in gameplyrs[0] else 1
if tar_item:
plyr_side = 0 if plyr_side else 1
sp = rec[KoS][0]
if sp in gameplyrs[plyr_side]:
# 85-86赛季开始由执法裁判记录
# 94-95赛季开始记录比赛总时长
import os
import pandas as pd
from bs4 import BeautifulSoup
from tqdm import tqdm
from util import getCode, writeToPickle, LoadPickle
import re
import time
for season in range(2020, 2021):
refs = LoadPickle('./data/refereeURLs.pickle')
res = [] # [gm, Referees, Attendance, Time of Game]
ss = '%d_%d' % (season-1, season)
# 赛季
print('=' * 50 + '\nstarting to record season %s' % ss)
seasondir = './data/seasons/%s/' % ss
for RoP in ['Regular', 'Playoff']:
tb = LoadPickle(seasondir + 'season%sSummary.pickle' % RoP)
for gm in tqdm(tb[1:]):
if gm[0] in ['195101140PHW']:
res.append([gm[0], '', 6665, ''])
elif gm[0] in ['197511260LAL']:
continue
else:
gameURL = 'https://www.basketball-reference.com/boxscores/%s.html' % gm[0]
gamePage = getCode(gameURL, 'UTF-8')
# s = time.time())
# q = gamePage.find('div', id='content').find_all('strong')
# print(time.time() - s)
def __init__(self, RoP):
self.dir = './data/seasons_RAT/'
self.plyrs = LoadPickle('./data/playerBasicInformation.pickle')
self.RoP = RoP
#!/usr/bin/python
# -*- coding:utf8 -*-
from klasses.Player import Player
from util import LoadPickle, writeToPickle
from klasses.stats_items import *
from tqdm import tqdm
import pandas as pd
# 计算球员赛季和生涯场均及总和并保存
pm2pn = LoadPickle('../data/playermark2playername.pickle')
for p in tqdm(list(pm2pn.keys())):
# print(p)
for RoP in ['regular', 'playoff']:
player = Player(p, RoP)
if player.exists and not isinstance(player.data, list):
res = []
for sss, ss in player.yieldSeasons():
sss = player.on_board_games(sss)
res += player.ave_and_sum(sss, type=2)
res[-2] = [ss] + res[-2]
res[-1] = [ss] + res[-1]
sss = player.on_board_games(player.data)
res += player.ave_and_sum(sss, type=2)
res[-2] = ['career'] + res[-2]
res[-1] = ['career'] + res[-1]
res = pd.DataFrame(res,
columns=['Season'] +
list(regular_items_en.keys() if RoP ==
'regular' else playoff_items_en.keys()))
# for col in res.columns:
# res[col] = res[col].astype('category')
import os
from tqdm import tqdm
import numpy as np
from klasses.Game import Game
from klasses.miscellaneous import MPTime
from util import writeToPickle, gameMarkToDir, LoadPickle, plus_minus
from splitLineups import splitLineups
regularOrPlayoffs = ['regular', 'playoff']
def inline(r):
line = r['R'][RoH]
return sorted(line, reverse=True)
lines_all = LoadPickle('./data/Lineups/anaSeason5Lineups.pickle')
for season in range(2020, 2021):
ss = '%d_%d' % (season, season + 1)
for i in range(2):
lines_all[i][ss] = {}
lines = {
} # {'tm1': {'line1': ['sum_time', '+/-'], 'line2': ['sum_time', '+/-'] ...}, 'tm2': ...}
gms = os.listdir('D:/sunyiwu/stat/data/seasons/%s/%s/' %
(ss, regularOrPlayoffs[i]))
for gm in tqdm(gms):
game = Game(gm, regularOrPlayoffs[i])
record = LoadPickle(gameMarkToDir(gm, regularOrPlayoffs[i], tp=3))
rot = game.rotation(record)
bxs, rot = game.replayer(record, rot)
et = '%d:00.0' % (48 + 5 * (game.quarters - 4))
for tmix, tm in enumerate(list(game.bxscr[0])): # 分别回溯两支球队
for ix, combs_i in enumerate(
pms1234): # 遍历1-4人的所有搭配，累加时间和正负值
for combs in combs_i:
combstr = ' '.join(sorted(combs))
if combstr not in lines1234[ix][tm]: # 每种组合初始化
lines1234[ix][tm][combstr] = [
MPTime('0:00.0'), 0
]
lines1234[ix][tm][combstr][0] += line5_tm[L][0]
lines1234[ix][tm][combstr][1] += line5_tm[L][1]
for ix in range(4):
for tm in lines1234[ix]:
lines_sorted = {}
for k in sorted(lines1234[ix][tm],
key=lines1234[ix][tm].__getitem__,
reverse=True): # 字典按值排序
lines_sorted[k] = lines1234[ix][tm][k]
lines_all[ix][i][ss][tm] = lines_sorted
for ix in range(4):
writeToPickle(
'D:/sunyiwu/stat/data/Lineups/anaSeason%dLineups.pickle' %
(ix + 1), lines_all[ix])
# for combs in lines_all[ix][0]['2019_2020']['LAL']:
# print(combs, lines_all[ix][0]['2019_2020']['LAL'][combs])
if __name__ == '__main__':
lines5 = LoadPickle('./data/Lineups/anaSeason5Lineups.pickle')
splitLineups(lines5)