2017年NBA球员数据分析

前言

原始数据可以通过我分享的资源获取
NBA–2017年数据表

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

获取数据

data = pd.read_csv("./data/nba_2017_nba_players_with_salary.csv")
data.head()
data.shape

(342, 38)

# 粗略观察数据的各个统计值
data.describe()

	Rk	AGE	MP	FG	FGA	FG%	3P	3PA	3P%	2P	...	GP	MPG	ORPM	DRPM	RPM	WINS_RPM	PIE	PACE	W	SALARY_MILLIONS
count	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000	320.000000	342.000000	...	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000	342.000000
mean	217.269006	26.444444	21.572515	3.483626	7.725439	0.446096	0.865789	2.440058	0.307016	2.620175	...	58.198830	21.572807	-0.676023	-0.005789	-0.681813	2.861725	9.186842	98.341053	28.950292	7.294006
std	136.403138	4.295686	8.804018	2.200872	4.646933	0.078992	0.780010	2.021716	0.134691	1.828714	...	22.282015	8.804121	2.063237	1.614293	2.522014	3.880914	3.585475	2.870091	14.603876	6.516326
min	1.000000	19.000000	2.200000	0.000000	0.800000	0.000000	0.000000	0.000000	0.000000	0.000000	...	2.000000	2.200000	-4.430000	-3.920000	-6.600000	-2.320000	-1.600000	87.460000	0.000000	0.030000
25%	100.250000	23.000000	15.025000	1.800000	4.225000	0.402250	0.200000	0.800000	0.280250	1.200000	...	43.500000	15.025000	-2.147500	-1.222500	-2.422500	0.102500	7.100000	96.850000	19.000000	2.185000
50%	205.500000	26.000000	21.650000	3.000000	6.700000	0.442000	0.700000	2.200000	0.340500	2.200000	...	66.000000	21.650000	-0.990000	-0.130000	-1.170000	1.410000	8.700000	98.205000	29.000000	4.920000
75%	327.750000	29.000000	29.075000	4.700000	10.400000	0.481000	1.400000	3.600000	0.373500	3.700000	...	76.000000	29.075000	0.257500	1.067500	0.865000	4.487500	10.900000	100.060000	39.000000	11.110000
max	482.000000	40.000000	37.800000	10.300000	24.000000	0.750000	4.100000	10.000000	1.000000	9.700000	...	82.000000	37.800000	7.270000	6.020000	8.420000	20.430000	23.000000	109.870000	66.000000	30.960000

8 rows × 35 columns

数据分析

数据相关性

data_cor = data.loc[:, ['RPM', 'AGE', 'SALARY_MILLIONS', 'ORB', 
                        'DRB', 'TRB','AST', 'STL',
                        'BLK', 'TOV', 'PF', 
                        'POINTS', 'GP', 'MPG', 'ORPM', 'DRPM']]

data_cor.head()

	RPM	AGE	SALARY_MILLIONS	ORB	DRB	TRB	AST	STL	BLK	TOV	PF	POINTS	GP	MPG	ORPM	DRPM
0	6.27	28	26.50	1.7	9.0	10.7	10.4	1.6	0.4	5.4	2.3	31.6	81	34.6	6.74	-0.47
1	4.81	27	26.50	1.2	7.0	8.1	11.2	1.5	0.5	5.7	2.7	29.1	81	36.4	6.38	-1.57
2	1.83	27	6.59	0.6	2.1	2.7	5.9	0.9	0.2	2.8	2.2	28.9	76	33.8	5.72	-3.89
3	4.35	23	22.12	2.3	9.5	11.8	2.1	1.3	2.2	2.4	2.2	28.0	75	36.1	0.45	3.90
4	4.20	26	16.96	2.1	8.9	11.0	4.6	1.4	1.3	3.7	3.9	27.0	72	34.2	3.56	0.64

# 获取两列数据之间的相关性
corr = data_cor.corr()
corr.head()

	RPM	AGE	SALARY_MILLIONS	ORB	DRB	TRB	AST	STL	BLK	TOV	PF	POINTS	GP	MPG	ORPM	DRPM
RPM	1.000000	0.175820	0.477542	0.388764	0.623515	0.587853	0.481971	0.599008	0.463097	0.492014	0.434226	0.604432	0.340810	0.549449	0.769822	0.578388
AGE	0.175820	1.000000	0.353312	-0.015752	0.088859	0.062064	0.114908	0.069892	-0.062917	0.030673	0.005512	0.031422	0.051863	0.099657	0.136177	0.100636
SALARY_MILLIONS	0.477542	0.353312	1.000000	0.264954	0.531569	0.482088	0.486159	0.446763	0.260288	0.536993	0.341512	0.635425	0.348093	0.594162	0.503682	0.102307
ORB	0.388764	-0.015752	0.264954	1.000000	0.731345	0.861103	-0.011632	0.169075	0.654265	0.274670	0.557957	0.284908	0.296975	0.342140	0.102113	0.476857
DRB	0.623515	0.088859	0.531569	0.731345	1.000000	0.976244	0.350786	0.485726	0.660733	0.598043	0.670708	0.648267	0.473376	0.684662	0.428433	0.426536

# 创建画布
plt.figure(figsize=(20, 8), dpi=100)

# 画出相关性热图
# param1: 数据
# param2; 正方形
# param3: 线宽
# param4: 显示值
sns.heatmap(corr, square=True, linewidths=0.1, annot=True)

基本数据排名分析

# 按照效率值排名

data.loc[:, ["PLAYER", "RPM",
             "AGE"]].sort_values(by="RPM", ascending=False).head()

	PLAYER	RPM	AGE
6	LeBron James	8.42	32
37	Chris Paul	7.92	31
8	Stephen Curry	7.41	28
120	Draymond Green	7.14	26
7	Kawhi Leonard	7.08	25

# 按照球员薪资排名

data.loc[:, ["PLAYER", "RPM", "AGE", 
             "SALARY_MILLIONS"]].sort_values(by="SALARY_MILLIONS", 
                                 ascending=False).head()

	PLAYER	RPM	AGE	SALARY_MILLIONS
6	LeBron James	8.42	32	30.96
25	Mike Conley	4.47	29	26.54
67	Al Horford	1.82	30	26.54
0	Russell Westbrook	6.27	28	26.50
1	James Harden	4.81	27	26.50

Seaborn常用的三个数据可视化方法

单变量：

# 利用seaborn中的distplot绘图来
#分别看一下球员薪水、效率值、年龄这三个信息的分布情况

# 设置显示风格
sns.set_style("darkgrid")

# 设置画布
plt.figure(figsize=(10, 10))

# 分割屏幕 -- 薪水
plt.subplot(3, 1, 1)
sns.distplot(data["SALARY_MILLIONS"])
plt.ylabel("salary")

# 分割屏幕 -- 效率值(真实贡献值)
plt.subplot(3, 1, 2)
sns.distplot(data["RPM"])
plt.ylabel("RPM")

# 分割屏幕 -- 年龄
plt.subplot(3, 1, 3)
sns.distplot(data["AGE"])
plt.ylabel("AGE")

双变量

sns.jointplot(data.AGE, data.SALARY_MILLIONS, kind="hex")

多变量

multi_data = data.loc[:, ['RPM','SALARY_MILLIONS','AGE','POINTS']]
multi_data.head()

	RPM	SALARY_MILLIONS	AGE	POINTS
0	6.27	26.50	28	31.6
1	4.81	26.50	27	29.1
2	1.83	6.59	27	28.9
3	4.35	22.12	23	28.0
4	4.20	16.96	26	27.0

# 多变量两两做一个表
sns.pairplot(multi_data)

衍生变量的一些可视化实践-以年龄为例

def age_cut(df):
    """年龄划分"""
    if df.AGE <= 24:
        return "young"
    elif df.AGE >= 30:
        return "old"
    else:
        return "best"

# 使用apply对年龄进行划分
# 函数作为一个对象，能作为参数传递给其它参数，并且能作为函数的返回值
# 循环data的每一个值, 带入age_cut求值, 得到的结果赋给age_cut colunm
data["age_cut"] = data.apply(lambda x:age_cut(x), axis=1)
data.head()

	Rk	PLAYER	POSITION	AGE	MP	FG	FGA	FG%	3P	3PA	...	MPG	ORPM	DRPM	RPM	WINS_RPM	PIE	PACE	W	SALARY_MILLIONS	age_cut
0	1	Russell Westbrook	PG	28	34.6	10.2	24.0	0.425	2.5	7.2	...	34.6	6.74	-0.47	6.27	17.34	23.0	102.31	46	26.50	best
1	2	James Harden	PG	27	36.4	8.3	18.9	0.440	3.2	9.3	...	36.4	6.38	-1.57	4.81	15.54	19.0	102.98	54	26.50	best
2	3	Isaiah Thomas	PG	27	33.8	9.0	19.4	0.463	3.2	8.5	...	33.8	5.72	-3.89	1.83	8.19	16.1	99.84	51	6.59	best
3	4	Anthony Davis	C	23	36.1	10.3	20.3	0.505	0.5	1.8	...	36.1	0.45	3.90	4.35	12.81	19.2	100.19	31	22.12	young
4	6	DeMarcus Cousins	C	26	34.2	9.0	19.9	0.452	1.8	5.0	...	34.2	3.56	0.64	4.20	11.26	17.8	97.11	30	16.96	best

5 rows × 39 columns

# 方便计数
data["cut"] = 1

data.loc[data.age_cut == "best"].SALARY_MILLIONS.head()

0    26.50
1    26.50
2     6.59
4    16.96
5    24.33
Name: SALARY_MILLIONS, dtype: float64

# 基于年龄段对球员薪水和效率值进行分析
sns.set_style("darkgrid")
plt.figure(figsize=(10,10), dpi=100)
plt.title("RPM and Salary")

x1 = data.loc[data.age_cut == "old"].SALARY_MILLIONS
y1 = data.loc[data.age_cut == "old"].RPM
plt.plot(x1, y1, "^")

x2 = data.loc[data.age_cut == "best"].SALARY_MILLIONS
y2 = data.loc[data.age_cut == "best"].RPM
plt.plot(x2, y2, "^")

x3 = data.loc[data.age_cut == "young"].SALARY_MILLIONS
y3 = data.loc[data.age_cut == "young"].RPM
plt.plot(x3, y3, ".")

# 取出多个变量画图
multi_data2 = data.loc[:, ['RPM','POINTS',
                           'TRB','AST','STL','BLK','age_cut']]
# 用hue来指定对应colunm所有取值中每一种对应的颜色
sns.pairplot(multi_data2, hue="age_cut")

球队数据分析

球队薪资排行

# .agg() 聚合方法 -- 字典
data.groupby(by="age_cut").agg({
     "SALARY_MILLIONS":np.max})

	SALARY_MILLIONS
age_cut
best	26.54
old	30.96
young	22.12

# 按照球队进行分类
data_team = data.groupby(by="TEAM").agg({
     "SALARY_MILLIONS":np.mean})

# 按照薪资进行分类, 降序排列
data_team.sort_values(by="SALARY_MILLIONS", 
                      ascending=False).head(10)

	SALARY_MILLIONS
TEAM
CLE	17.095000
HOU	13.432000
GS	12.701429
ORL/TOR	11.125000
POR	9.730000
WSH	9.628889
ORL	9.490000
MIL/CHA	9.425000
SA	9.347273
NO/SAC	8.970000

# 按照分球队分年龄段，上榜球员降序排列，
# 如上榜球员数相同，则按效率值降序排列。
data_rpm = data.groupby(by=["TEAM",
           "age_cut"]).agg({
     "SALARY_MILLIONS": np.mean,
                            "RPM": np.mean, "PLAYER": np.size})
data_rpm.sort_values(by=["PLAYER", "RPM"], ascending=False).head()
data_rpm.head()

		SALARY_MILLIONS	RPM	PLAYER
TEAM	age_cut
ATL	best	4.678000	-1.768000	5
	old	12.775000	0.982500	4
	young	1.926667	-3.076667	3
ATL/CLE	old	5.040000	-2.485000	2
ATL/PHI/OKC	best	8.400000	1.720000	1

按照球队综合实力排名

data_rpm2 = data.groupby(by=['TEAM'],
                         as_index=False).agg({
     'SALARY_MILLIONS': np.mean,
                                                          'RPM': np.mean,
                                                          'PLAYER': np.size,
                                                          'POINTS': np.mean,
                                                          'eFG%': np.mean,
                                                          'MPG': np.mean,
                                                          'AGE': np.mean})
data_rpm2.head()

	TEAM	SALARY_MILLIONS	RPM	PLAYER	POINTS	eFG%	MPG	AGE
0	ATL	6.689167	-1.178333	12	7.416667	0.442667	18.541667	27.000000
1	ATL/CLE	5.040000	-2.485000	2	7.650000	0.582000	21.050000	35.500000
2	ATL/PHI/OKC	8.400000	1.720000	1	13.100000	0.511000	26.100000	29.000000
3	BKN	5.704545	-1.224545	11	9.045455	0.487273	20.227273	27.636364
4	BKN/WSH	4.910000	-4.045000	2	8.150000	0.470000	17.350000	27.000000

# 按照效率值降序排列
data_rpm2.sort_values(by="RPM", ascending=False).head()

	TEAM	SALARY_MILLIONS	RPM	PLAYER	POINTS	eFG%	MPG	AGE
18	GS	12.701429	3.478571	7	14.528571	0.575143	26.700000	28.714286
9	CLE	17.095000	2.566667	6	15.883333	0.555833	29.766667	28.000000
2	ATL/PHI/OKC	8.400000	1.720000	1	13.100000	0.511000	26.100000	29.000000
20	HOU	13.432000	1.582000	5	15.420000	0.534600	29.980000	27.200000
44	SA	9.347273	0.901818	11	9.818182	0.524182	21.472727	29.545455

利用箱线图和小提琴图进行数据分析

# 筛选数据
data.TEAM.isin(['GS', 'CLE', 'SA', 'LAC',
                'OKC', 'UTAH', 'CHA', 'TOR', 'NO', 'BOS']).head()

0     True
1    False
2     True
3     True
4    False
Name: TEAM, dtype: bool

# 箱线图

#设置图片背景
sns.set_style("whitegrid")
plt.figure(figsize=(20, 10))

# 获取需要的数据
data_team2 = data[data.TEAM.isin(['GS', 'CLE', 'SA', 'LAC',
                                  'OKC', 'UTAH', 'CHA', 
                                  'TOR', 'NO', 'BOS'])]

# 进行相应的绘图\

# 年薪
plt.subplot(3,1,1)
sns.boxplot(x="TEAM", y="SALARY_MILLIONS", data = data_team2)

# 年龄
plt.subplot(3,1,2)
sns.boxplot(x="TEAM", y="AGE", data = data_team2)

# 场均上场时间
plt.subplot(3,1,3)
sns.boxplot(x="TEAM", y="MPG", data = data_team2)

# 绘制小提琴图

# 设置图背景
sns.set_style("whitegrid")
plt.figure(figsize=(20, 10))

# 三分命中率
plt.subplot(3,1,1)
sns.violinplot(x="TEAM", y="3P%", data=data_team2)

# 有效命中率
plt.subplot(3,1,2)
sns.violinplot(x="TEAM", y="eFG%", data=data_team2)

# 得分
plt.subplot(3,1,3)
sns.violinplot(x="TEAM", y="POINTS", data=data_team2)