绝地求生预测获胜排名百分位数

import numpy as np
import pandas as pd

# 展示
import matplotlib.pyplot as plt
import seaborn as sns
from pdpbox import pdp
# Sklearn
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import train_test_split
from IPython.display import display
from sklearn import metrics

1、数据介绍

• DBNOs - Number of enemy players knocked.
• assists - Number of enemy players this player damaged that were killed by teammates.
• boosts - Number of boost items used.
• damageDealt - Total damage dealt. Note: Self inflicted damage is subtracted.
• headshotKills - Number of enemy players killed with headshots.
• heals - Number of healing items used.
• Id - Player’s Id
• killPlace - Ranking in match of number of enemy players killed.
• killPoints - Kills-based external ranking of player. (Think of this as an Elo ranking where only kills matter.) If there is a value other - than -1 in rankPoints, then any 0 in killPoints should be treated as a “None”.
• killStreaks - Max number of enemy players killed in a short amount of time.
• kills - Number of enemy players killed.
• longestKill - Longest distance between player and player killed at time of death. This may be misleading, as downing a player and driving away may lead to a large longestKill stat.
• matchDuration - Duration of match in seconds.
• matchId - ID to identify match. There are no matches that are in both the training and testing set.
• matchType - String identifying the game mode that the data comes from. The standard modes are “solo”, “duo”, “squad”, “solo-fpp”, “duo-fpp”, and “squad-fpp”; other modes are from events or custom matches.
• rankPoints - Elo-like ranking of player. This ranking is inconsistent and is being deprecated in the API’s next version, so use with caution. Value of -1 takes place of “None”.
• revives - Number of times this player revived teammates.
• rideDistance - Total distance traveled in vehicles measured in meters.
• roadKills - Number of kills while in a vehicle.
• swimDistance - Total distance traveled by swimming measured in meters.
• teamKills - Number of times this player killed a teammate.
• vehicleDestroys - Number of vehicles destroyed.
• walkDistance - Total distance traveled on foot measured in meters.
• weaponsAcquired - Number of weapons picked up.
• winPoints - Win-based external ranking of player. (Think of this as an Elo ranking where only winning matters.) If there is a value other than -1 in rankPoints, then any 0 in winPoints should be treated as a “None”.
• groupId - ID to identify a group within a match. If the same group of players plays in different matches, they will have a different groupId each time.
• numGroups - Number of groups we have data for in the match.
• maxPlace - Worst placement we have data for in the match. This may not match with numGroups, as sometimes the data skips over placements.
• winPlacePerc - The target of prediction. This is a percentile winning placement, where 1 corresponds to 1st place, and 0 corresponds to last place in the match. It is calculated off of maxPlace, not numGroups, so it is possible to have missing chunks in a match.

train = pd.read_csv('train_V2.csv')
test = pd.read_csv('test_V2.csv')

train.head()

	Id	groupId	matchId	assists	boosts	damageDealt	DBNOs	headshotKills	heals	killPlace	...	revives	rideDistance	roadKills	swimDistance	teamKills	vehicleDestroys	walkDistance	weaponsAcquired	winPoints	winPlacePerc
0	7f96b2f878858a	4d4b580de459be	a10357fd1a4a91	0	0	0.00	0	0	0	60	...	0	0.0000	0	0.00	0	0	244.80	1	1466	0.4444
1	eef90569b9d03c	684d5656442f9e	aeb375fc57110c	0	0	91.47	0	0	0	57	...	0	0.0045	0	11.04	0	0	1434.00	5	0	0.6400
2	1eaf90ac73de72	6a4a42c3245a74	110163d8bb94ae	1	0	68.00	0	0	0	47	...	0	0.0000	0	0.00	0	0	161.80	2	0	0.7755
3	4616d365dd2853	a930a9c79cd721	f1f1f4ef412d7e	0	0	32.90	0	0	0	75	...	0	0.0000	0	0.00	0	0	202.70	3	0	0.1667
4	315c96c26c9aac	de04010b3458dd	6dc8ff871e21e6	0	0	100.00	0	0	0	45	...	0	0.0000	0	0.00	0	0	49.75	2	0	0.1875

5 rows × 29 columns

2、简单清洗数据

2.1 清洗空数据

train[train['winPlacePerc'].isnull()]

	Id	groupId	matchId	assists	boosts	damageDealt	DBNOs	headshotKills	heals	killPlace	...	revives	rideDistance	roadKills	swimDistance	teamKills	vehicleDestroys	walkDistance	weaponsAcquired	winPoints	winPlacePerc
2744604	f70c74418bb064	12dfbede33f92b	224a123c53e008	0	0	0.0	0	0	0	1	...	0	0.0	0	0.0	0	0	0.0	0	0	NaN

1 rows × 29 columns

train.drop(2744604,inplace=True)
train[train['winPlacePerc'].isnull()]

	Id	groupId	matchId	assists	boosts	damageDealt	DBNOs	headshotKills	heals	killPlace	...	revives	rideDistance	roadKills	swimDistance	teamKills	vehicleDestroys	walkDistance	weaponsAcquired	winPoints	winPlacePerc

0 rows × 29 columns

2.2 去除每组人数不同对战绩的影响

train['playersJoined'] = train.groupby('matchId')['matchId'].transform('count')
plt.figure(figsize=(15,10))
sns.countplot(train[train['playersJoined']>=75]['playersJoined']) 
'''sns.countplot 是 Seaborn 库中的一个函数，用于绘制分类变量的计数直方图。它主要用于统计每个类别中数据出现的次数，并将结果以直方图的形式可视化展示出来。
这个函数的调用形式通常是 sns.countplot(x='variable', data=data)，其中 x 是分类变量的名称，data 是包含数据的 DataFrame 或其他数据结构。'''
plt.show()

train.head()

	Id	groupId	matchId	assists	boosts	damageDealt	DBNOs	headshotKills	heals	killPlace	...	rideDistance	roadKills	swimDistance	teamKills	vehicleDestroys	walkDistance	weaponsAcquired	winPoints	winPlacePerc	playersJoined
0	7f96b2f878858a	4d4b580de459be	a10357fd1a4a91	0	0	0.00	0	0	0	60	...	0.0000	0	0.00	0	0	244.80	1	1466	0.4444	96
1	eef90569b9d03c	684d5656442f9e	aeb375fc57110c	0	0	91.47	0	0	0	57	...	0.0045	0	11.04	0	0	1434.00	5	0	0.6400	91
2	1eaf90ac73de72	6a4a42c3245a74	110163d8bb94ae	1	0	68.00	0	0	0	47	...	0.0000	0	0.00	0	0	161.80	2	0	0.7755	98
3	4616d365dd2853	a930a9c79cd721	f1f1f4ef412d7e	0	0	32.90	0	0	0	75	...	0.0000	0	0.00	0	0	202.70	3	0	0.1667	91
4	315c96c26c9aac	de04010b3458dd	6dc8ff871e21e6	0	0	100.00	0	0	0	45	...	0.0000	0	0.00	0	0	49.75	2	0	0.1875	97

5 rows × 30 columns

train['killsNorm'] = train['kills']*((100-train['playersJoined'])/100+1)
train['damageDealtNorm'] = train['damageDealt']*((100-train['playersJoined'])/100+1)
train['matchDurationNorm'] = train['matchDuration']*((100-train['playersJoined'])/100+1)
to_show = ['Id', 'kills','killsNorm','damageDealt', 'damageDealtNorm', 'matchDuration', 'matchDurationNorm']
train[to_show][:11]

	Id	kills	killsNorm	damageDealt	damageDealtNorm	matchDuration	matchDurationNorm
0	7f96b2f878858a	0	0.00	0.000	0.00000	1306	1358.24
1	eef90569b9d03c	0	0.00	91.470	99.70230	1777	1936.93
2	1eaf90ac73de72	0	0.00	68.000	69.36000	1318	1344.36
3	4616d365dd2853	0	0.00	32.900	35.86100	1436	1565.24
4	315c96c26c9aac	1	1.03	100.000	103.00000	1424	1466.72
5	ff79c12f326506	1	1.05	100.000	105.00000	1395	1464.75
6	95959be0e21ca3	0	0.00	0.000	0.00000	1316	1355.48
7	311b84c6ff4390	0	0.00	8.538	8.87952	1967	2045.68
8	1a68204ccf9891	0	0.00	51.600	53.14800	1375	1416.25
9	e5bb5a43587253	0	0.00	37.270	38.38810	1930	1987.90
10	2b574d43972813	0	0.00	28.380	28.66380	1811	1829.11

3、对于开挂的，应当给他们剔除掉

3.1 第一种外挂：没有步数便击杀人头的

train['totalDistance'] = train['rideDistance'] + train['walkDistance'] + train['swimDistance']
train['killWithoutMoving'] = ((train['kills']>0)&(train['totalDistance']==0))
#train[train['killWithoutMoving']==True].shape
train.drop(train[train['killWithoutMoving']==True].index,inplace=True)

3.2 第二种外挂：坐在车上便能击杀人头的

train[train['roadKills']>10].shape

(4, 35)

train.drop(train[train['roadKills']>10].index,inplace=True)

3.3 第三种外挂：击杀人头数高的离谱的

plt.figure(figsize=(15,8))
sns.countplot(data=train,x=train['kills']).set_title('Kills')
plt.show()

train[train['kills']>30].shape
train.drop(train[train['kills']>30].index,inplace=True)

3.4 第四种外挂：爆头击杀率很高的且击杀人头数多的

'''
headshotKills - Number of enemy players killed with headshots.#爆头击杀，直中要害
kills - Number of enemy players killed.#被击杀
'''
train['headshot_rate'] = train['headshotKills']/train['kills']
train['headshot_rate'] = train['headshot_rate'].fillna(0)

plt.figure(figsize=(15,10))
sns.distplot(train['headshot_rate'],bins=10)
plt.show()

train[train['headshot_rate']==1].shape

(253959, 36)

train[  (train['headshot_rate']==1)  & (train['kills']==1)  ].shape

(218433, 36)

train.drop(train[  (train['headshot_rate']==1)  & (train['kills']>=5)  ].index,inplace=True)

4、使用categorical变量，减少系统占用内存

train['matchId'] = train['matchId'].astype('category')
train['groupId'] = train['groupId'].astype('category')

5、查看不同模式下击杀人头数与胜率关系

train.drop(columns=['Id'],inplace=True)

5.1 单排，双排，四排

solos=train[train['numGroups']>50]
duos=train[(train['numGroups']>25)&(train['numGroups']<=50)]
squads=train[train['numGroups']<=25]

len(solos)/len(train)

0.15947449945499398

len(duos)/len(train)

0.7412968331156684

f,ax =plt.subplots(figsize=(20,10))
sns.pointplot(x='kills',y='winPlacePerc',data=solos,color='black',alpha=0.8)
sns.pointplot(x='kills',y='winPlacePerc',data=duos,color='red',alpha=0.8)
sns.pointplot(x='kills',y='winPlacePerc',data=squads,color='blue',alpha=0.8)
plt.text(25,0.5,'Solos',color='red')
plt.grid()
plt.show()

6 、热力图

k = 5
f,ax =plt.subplots(figsize=(12,12))
temp_train=train.drop(columns=['groupId','matchId','matchType']) 

cols = temp_train.corr().nlargest(k,'winPlacePerc')['winPlacePerc'].index
#train.corr().nlargest(k,'winPlacePerc')['winPlacePerc']返回一个与'winPlacePerc'相关性最大的5分变量以及相关系数
# a 0.09
# b 0.01
# c 0.98
#此时相关系数的名称为'winPlacePerc'
cm = np.corrcoef(temp_train[cols].values.T)
sns.heatmap(cm,annot=True,linewidths=0.5,fmt='.1f',ax=ax,yticklabels=cols.values,xticklabels=cols.values)
plt.show()

7、建模

sample = 500000
df_sample = train.sample(sample)
df_sample.drop(columns = ['groupId','matchId','matchType'],inplace=True)

df = df_sample.drop(columns=['winPlacePerc'])
y = df_sample['winPlacePerc']

7.1 将训练集分为训练部分与测试部分

X_train,X_valid,y_train,y_valid = train_test_split(df,y,random_state=1)

7.2 训练测试集

def print_score(m):
    res= ['mae train',mean_absolute_error(m.predict(X_train),y_train),
         'mae val',mean_absolute_error(m.predict(X_valid),y_valid)]
    print (res)

from sklearn.metrics import mean_absolute_error
m1 = RandomForestRegressor(n_estimators=50,n_jobs=-1)
m1.fit(X_train,y_train)
print_score(m1)

['mae train', 0.02192784867733333, 'mae val', 0.05833728120000001]

m1.feature_importances_

array([1.39260048e-03, 5.35238158e-03, 3.30978196e-03, 2.28083086e-03,
       4.01083836e-04, 2.68734056e-03, 1.81916824e-01, 2.36602622e-03,
       3.11364071e-03, 2.75269723e-03, 6.13427821e-03, 9.33855372e-03,
       5.53881534e-03, 1.20290463e-02, 4.37883164e-03, 8.08006977e-04,
       1.91166905e-03, 4.32997862e-05, 7.26137113e-04, 2.31430733e-04,
       8.58912251e-05, 6.76056200e-01, 3.81913691e-03, 2.60518373e-03,
       1.84332720e-02, 8.68569187e-03, 3.87405597e-03, 1.14654252e-02,
       2.75947316e-02, 0.00000000e+00, 6.67135806e-04])

def rf_feat_importance(m,df):
    return pd.DataFrame({'cols':df.columns,'imp':m.feature_importances_}).sort_values('imp',ascending=False)

rf_feat_importance(m1,df)

	cols	imp
21	walkDistance	0.676056
6	killPlace	0.181917
28	totalDistance	0.027595
24	playersJoined	0.018433
13	numGroups	0.012029
27	matchDurationNorm	0.011465
11	matchDuration	0.009339
25	killsNorm	0.008686
10	longestKill	0.006134
12	maxPlace	0.005539
1	boosts	0.005352
14	rankPoints	0.004379
26	damageDealtNorm	0.003874
22	weaponsAcquired	0.003819
2	damageDealt	0.003310
8	kills	0.003114
9	killStreaks	0.002753
5	heals	0.002687
23	winPoints	0.002605
7	killPoints	0.002366
3	DBNOs	0.002281
16	rideDistance	0.001912
0	assists	0.001393
15	revives	0.000808
18	swimDistance	0.000726
30	headshot_rate	0.000667
4	headshotKills	0.000401
19	teamKills	0.000231
20	vehicleDestroys	0.000086
17	roadKills	0.000043
29	killWithoutMoving	0.000000

rf_feat_importance(m1,df)[:10].plot('cols','imp',figsize=(14,6),kind='barh')
plt.show()

fi=rf_feat_importance(m1,df)

to_keep = fi[fi.imp>0.01].cols
to_keep

21         walkDistance
6             killPlace
28        totalDistance
24        playersJoined
13            numGroups
27    matchDurationNorm
Name: cols, dtype: object

7.3 用取出的重要指标计算训练集误差

X_train,X_valid = X_train[to_keep],X_valid[to_keep]

m2 = RandomForestRegressor(n_estimators=50,n_jobs=-1)
m2.fit(X_train,y_train)
print_score(m2)

['mae train', 0.02192277282133332, 'mae val', 0.05834604831999999]

8、预测测试集

temp_test=test.copy()

temp_test.head()

	Id	groupId	matchId	assists	boosts	damageDealt	DBNOs	headshotKills	heals	killPlace	...	rankPoints	revives	rideDistance	roadKills	swimDistance	teamKills	vehicleDestroys	walkDistance	weaponsAcquired	winPoints
0	9329eb41e215eb	676b23c24e70d6	45b576ab7daa7f	0	0	51.46	0	0	0	73	...	1500	0	0.0	0	0.0	0	0	588.0	1	0
1	639bd0dcd7bda8	430933124148dd	42a9a0b906c928	0	4	179.10	0	0	2	11	...	1503	2	4669.0	0	0.0	0	0	2017.0	6	0
2	63d5c8ef8dfe91	0b45f5db20ba99	87e7e4477a048e	1	0	23.40	0	0	4	49	...	1565	0	0.0	0	0.0	0	0	787.8	4	0
3	cf5b81422591d1	b7497dbdc77f4a	1b9a94f1af67f1	0	0	65.52	0	0	0	54	...	1465	0	0.0	0	0.0	0	0	1812.0	3	0
4	ee6a295187ba21	6604ce20a1d230	40754a93016066	0	4	330.20	1	2	1	7	...	1480	1	0.0	0	0.0	0	0	2963.0	4	0

5 rows × 28 columns

temp_test只有三列是自带的，其他列均需要处理下
21 walkDistance 25
6 killPlace 9
28 totalDistance
24 playersJoined
13 numGroups 17
27 matchDurationNorm

temp_test['totalDistance']=temp_test['rideDistance'] + temp_test['walkDistance'] + temp_test['swimDistance']
temp_test['playersJoined'] = temp_test.groupby('matchId')['matchId'].transform('count')
temp_test['matchDurationNorm'] = temp_test['matchDuration']*((100-temp_test['playersJoined'])/100+1)

21 walkDistance 25

6 killPlace 9

28 totalDistance 28

24 playersJoined 29

13 numGroups 17

27 matchDurationNorm 30

#训练一个只有特征和指标的模型
X=X_train[['walkDistance','killPlace','totalDistance','playersJoined','numGroups','matchDurationNorm']]
fratures_model = RandomForestRegressor(n_estimators=50,n_jobs=-1)

fratures_model.fit(X,y_train)

temp_test['winPlacePerc']=fratures_model.predict(temp_test[['walkDistance','killPlace','totalDistance','playersJoined','numGroups','matchDurationNorm']])

temp_test.head()

	Id	groupId	matchId	assists	boosts	damageDealt	DBNOs	headshotKills	heals	killPlace	...	swimDistance	teamKills	vehicleDestroys	walkDistance	weaponsAcquired	winPoints	totalDistance	playersJoined	matchDurationNorm	winPlacePerc
0	9329eb41e215eb	676b23c24e70d6	45b576ab7daa7f	0	0	51.46	0	0	0	73	...	0.0	0	0	588.0	1	0	588.0	92	2034.72	0.183950
1	639bd0dcd7bda8	430933124148dd	42a9a0b906c928	0	4	179.10	0	0	2	11	...	0.0	0	0	2017.0	6	0	6686.0	96	1883.44	0.858322
2	63d5c8ef8dfe91	0b45f5db20ba99	87e7e4477a048e	1	0	23.40	0	0	4	49	...	0.0	0	0	787.8	4	0	787.8	94	1900.58	0.718864
3	cf5b81422591d1	b7497dbdc77f4a	1b9a94f1af67f1	0	0	65.52	0	0	0	54	...	0.0	0	0	1812.0	3	0	1812.0	89	2035.74	0.520244
4	ee6a295187ba21	6604ce20a1d230	40754a93016066	0	4	330.20	1	2	1	7	...	0.0	0	0	2963.0	4	0	2963.0	95	1392.30	0.881686

5 rows × 32 columns