機器學習（八）：nba數據分析小案例

注：基於實驗樓一個小項目

數據下載地址：

http://labfile.oss.aliyuncs.com/courses/782/data.zip

代碼如下：

import pandas as pd
import math
import csv
import random
import numpy as np
from sklearn import linear_model
from sklearn.model_selection import cross_val_score
base_elo = 1600
team_elos = {}
team_stats = {}
X = []
y = []
#初始化數據，從T，O，M表格中讀取數據，取出一些無關數據並將這三個表格通過team樹形列進行連接：
#根據每個隊伍的Miscellaneous Opponent，Team統計數據csv文件進行初始化
def initialize_data(Mstat,Ostat,Tstat):
    new_Mstat = Mstat.drop(['Rk','Arena'],axis=1)
    new_Ostat = Ostat.drop(['Rk',"G",'MP'],axis=1)
    new_Tstat = Tstat.drop(['Rk',"G",'MP'],axis=1)
    team_stats1 = pd.merge(new_Mstat,new_Ostat,how='left',on='Team')
    team_stats1 = pd.merge(team_stats1,new_Tstat,how='left',on='Team')
    return team_stats1.set_index('Team',inplace=False,drop=True)
def get_elo(team):
    try:
        return team_elos[team]
    except:
        team_elos[team] = base_elo
    return team_elos[team]
def calc_elo(win_team,lose_team):
    winner_rank = get_elo(win_team)
    loser_rank = get_elo(lose_team)
    #根據Logistic Distribution計算 PK 雙方（A和B）對各自的勝率期望值計算公式
    rank_diff = winner_rank - loser_rank
    exp = (rank_diff *-1)/400
    odds  = 1/(1+math.pow(10,exp))
    #根據rank界別修改k值
    if winner_rank < 2100:
        k = 32
    elif winner_rank >=2100 and winner_rank <2400:
        k = 24
    else:
        k=16
    #更新rank數值
    new_winner_rank = round(winner_rank+(k*(1-odds)))
    new_loser_rank = round(loser_rank+(k*(0-odds)))
    return new_winner_rank,new_loser_rank

#基於統計好的數據，給每隻隊伍的eloscore計算結果，建立對應15-16年數據集，我們認爲主場作戰的隊伍更有優勢，因此會給主場隊伍加上100分
def build_dataSet(all_data):
    print("Building data set..")
    X = []
    skip = 0
    for index,row in all_data.iterrows():
        Wteam = row['WTeam']
        Lteam = row['LTeam']
        #獲取最初的elo或者每個隊伍最初的elo值
        team1_elo = get_elo(Wteam)
        team2_elo = get_elo(Lteam)
        #給主場比賽隊伍加上100的elo值
        if row['WLoc'] == 'H':
            team1_elo += 100
        else:
            team2_elo += 100
        #把elo當成評價每個隊伍的第一個特徵值
        team1_features = [team1_elo]
        team2_features = [team2_elo]
        # 添加我們從basketball reference.com獲得的每個隊伍的統計信息
        for key,value in team_stats.loc[Wteam].iteritems():
            team1_features.append(value)
        for key,value in team_stats.loc[Lteam].iteritems():
            team2_features.append(value)
        # 將兩支隊伍的特徵值隨機的分配在每場比賽數據的左右兩側
        # 並將對應的0/1賦給y值
        if random.random() > 0.5:
            X.append(team1_features+team2_features)
            y.append(0)
        else:
            X.append(team2_features+team1_features)
            y.append(1)
        if skip ==0:
            print('X',X)
            skip = 1
        new_winner_rank,new_loser_rank = calc_elo(Wteam,Lteam)
        team_elos[Wteam] = new_winner_rank
        team_elos[Lteam] = new_loser_rank
    return np.nan_to_num(X),y
#最終利用訓練好的模型在 16~17 年的常規賽數據中進行預測
def predict_winner(team_1, team_2, model):
    features = []
    # team 1，客場隊伍
    features.append(get_elo(team_1))
    for key, value in team_stats.loc[team_1].iteritems():
        features.append(value)
    # team 2，主場隊伍
    features.append(get_elo(team_2) + 100)
    for key, value in team_stats.loc[team_2].iteritems():
        features.append(value)
    features = np.nan_to_num(features)
    return model.predict_proba([features])
#最終在 main 函數中調用這些數據處理函數，使用 sklearn 的Logistic Regression方法建立迴歸模型
if __name__=='__main__':
    folder = 'data'
    Mstat = pd.read_csv(folder + '/15-16Miscellaneous_Stat.csv')
    Ostat = pd.read_csv(folder + '/15-16Opponent_Per_Game_Stat.csv')
    Tstat = pd.read_csv(folder + '/15-16Team_Per_Game_Stat.csv')
    team_stats = initialize_data(Mstat, Ostat, Tstat)
    result_data = pd.read_csv(folder + '/2015-2016_result.csv')
    X, y = build_dataSet(result_data)
    #訓練網絡模型
    print("Fitting on %d game samples.." % len(X))
    model = linear_model.LogisticRegression()
    model.fit(X,y)
    print("Doing cross-validation..")
    cross_val_score(model,X,y,cv = 10,scoring='accuracy',n_jobs=-1).mean()
    print(model)
    print('Predicting on new schedule..')
    schedule1617 = pd.read_csv(folder + '/16-17Schedule.csv')
    result = []
    for index, row in schedule1617.iterrows():
        team1 = row['Vteam']
        team2 = row['Hteam']
        pred = predict_winner(team1, team2, model)
        prob = pred[0][0]
        if prob > 0.5:
            winner = team1
            loser = team2
            result.append([winner, loser, prob])
        else:
            winner = team2
            loser = team1
            result.append([winner, loser, 1 - prob])
    with open('16-17Result.csv', 'w') as f:
        writer = csv.writer(f)
        writer.writerow(['win', 'lose', 'probability'])
        writer.writerows(result)
        print('done.')

查看最後得到的模型:

解釋：

函數 initialize_data是對讀取來的csv文件中的數據進行初始化，刪除無關數據，並通過team列將讀取來的csv數值進行鏈接

函數get_elo 是初始化以下elo數值，方面後面的邏輯迴歸中的勝率期望值計算公試計算，elo也是這裏面的概念，初步都初始化1600

函數calc_elo就是勝率期望計算公式計算了，得到最後的elo等級分

函數build_dataSet調用上面的相關函數，得到最後的elo分值，得到最後的數值後使用 sklearn 的Logistic Regression方法建立迴歸模型。得到上面的模型。

然後predict_winner函數就是針對一場新的比賽進行預測了。預測結果放在了16-17Result.csv文件中