NBA kaggle analysis

本文参考，并将其延伸至2018-2020 赛季。全文用rmd 完成，ggplot2为主要做图工具。主要想研究下各个球队和伤病之间的关系

https://www.kaggle.com/jaseziv83/extensive-nba-injuries-deep-dive-eda/comments

数据下载

https://www.kaggle.com/ghopkins/nba-injuries-2010-2018

数据清理

返回头部数据看下格式

注：这里的Acquired 和 Relinquished是呈现互补关系。若Acquired非空，则Relinquished 空

最后的记录是到2020-10-2.

发现存在很多空的列表，看下个数（这里有一个坑就是这里的缺失值不是NA，而是“”。）

1	print(colSums(df == ""))

再看一下总体数据的size

1	glimpse(df)

Team 列查看

1	df %>% dplyr::filter(Team == "")

这两例情况其实都是受伤了，被球队裁掉了

这里以大宝贝为例

所以我们可以大胆的排除掉这两种情况

Acquired列查看

df %>%
  filter(Acquired != "") %>%
  count(Notes) %>%
  arrange(desc(n))

发现这是一些恢复的数据，而我们的目的是找到一些球队和伤病的问题，所以我们可以单纯的不考虑这些。正如我前面提到的那样，Acquired 和 Relinquished 是互补的，所以我们可以如下处理数据

df_exc <- df %>%
  filter(Acquired == "") %>% # 排除Acquired==““
  filter(Team != "") %>% # 排除Team==““
  select(-Acquired) "直接删掉Acquired这一个变量"
print(colSums(df_exc == "")) # 检查一下还有没有多的缺失值

还有四个漏网之鱼，也处理掉

1
2
3

### Exclude some missing values in Relinquished variable
df <- df_exc %>% filter(Relinquished != "")
print(colSums(df == ""))

到目前为止，缺失值已经处理完毕了

以时间为单位来进行处理（转换成Date格式）

1
2
3

df$year <- format(as.Date(df$Date), "%Y")
df$yr_mo <- format(as.Date(df$Date), "%Y-%m")
df$month <- month(as.Date(df$Date), TRUE)

看一下这十年究竟谁伤病情况（包含轮休）的最多

df %>%
  count(Team) %>%
  ggplot(aes(x = reorder(Team, n), y = n)) +
  geom_col(fill = "midnightblue", color = "white") +
  coord_flip() +
  theme_bw() +
  labs(x = "Team", y = "Number of injury events", title = "Which team has the most injury events") +
  theme(panel.grid.major.y = element_blank(), panel.background = element_rect(fill = "whitesmoke"), plot.title = element_text(hjust = 0.5))

前三名分别是雄鹿，马刺和火箭。

由于联盟在2018 年发布了新的政策，对非伤轮休的球队会处以罚款，之前的记录可能有一些水分。所以我又做了2018年之后的数据

df %>%
  filter(yr_mo > "2018-09") %>%
  count(Team) %>%
  ggplot(aes(x = reorder(Team, n), y = n)) +
  geom_col(fill = "midnightblue", color = "white") +
  coord_flip() +
  theme_bw() +
  labs(x = "Team", y = "Number of injury events", title = "Which team has the most injury events since 2018-2019 season") +
  theme(panel.grid.major.y = element_blank(), panel.background = element_rect(fill = "whitesmoke"), plot.title = element_text(hjust = 0.5))

马刺在本次赛季失去效力了十七年的后卫托尼·帕克，他在休赛期间选择与夏洛特黄蜂签约，而杰弗瑞·罗德曼吉、凯尔·安德森以及布兰登·保罗等重要角色球员也先后离队。这也是自2002–03赛季以来第一次没有马努·吉诺比利的赛季，他在2018年8月27日选择对外宣布退役，因此这也标志着马刺的“三巨头时代”至此已经彻底结束。2018年7月18日，马刺管理层将科怀·伦纳德和丹尼·格林交易到多伦多猛龙，以换取德玛尔·德罗赞、雅各·伯尔特和一枚受保护的2019年首轮选秀签。

新鲜血液的加入以及整体阵容的年轻化，马刺在2018-2019赛季几乎全员健康，受伤病影响最小。

此时受伤病影响最严重的前三只球队分别是凯尔特人，雄鹿和勇士。

十年内哪位球员受伤病影响（含轮休）最多

df %>%
  #filter(yr_mo > "2018-09" & yr_mo< "2019-09") %>%
  count(Relinquished) %>%
  arrange(desc(n)) %>%
  top_n(n = 20) %>%
  ggplot(aes(x = reorder(Relinquished, n), y = n)) +
  geom_col(fill = "midnightblue", color = "white") +
  coord_flip() +
  theme_bw() +
  labs(x = "Player", y = "Number of injury events on players", title = "Which player has the most injury events") +
  theme(panel.grid.major.y = element_blank(), panel.background = element_rect(fill = "whitesmoke"), plot.title = element_text(hjust = 0.5))

乐福，戈登和罗斯是近十年来受伤病影响最大的（轮休也是预防伤病）

每年nba球员受伤病伤病影响的数量

df %>%
count(year, Relinquished) %>%
ggplot(aes(x = n)) +
geom_density(color = "midnightblue", fill = "white") +
facet_wrap(~year, scales = "free_y") +
theme_bw() +
labs(y = "", x = "Number of events", title = "Number of events per player", subtitle = "Note: 2010 & 2020 not a full year") +
theme(plot.title = element_text(hjust = 0.5), plot.subtitle = element_text(hjust = 0.5), panel.grid.major.y = element_blank(), panel.background = element_rect(fill = "whitesmoke"))

每年的分布趋势都大致相同

NBA 球员每年受伤数量汇总（折线图）

df %>%
  count(year) %>%
  ggplot(aes(x = year, y = n, group = 1)) +
  ylim(0, 2700) +
  geom_line(size = 1, color = "darkred") +
  geom_point(size = 2, color = "midnightblue") +
  geom_text(color = "midnightblue", aes(label = n, y = n + 100)) +
  theme_classic() +
  labs(title = "Tracking injury events over time", subtitle = "Note: 2010 & 2018 not a full year", x = "Calendar Year", y = "Number events") +
  theme(panel.grid.major.y = element_blank(), panel.background = element_rect(fill = "whitesmoke"), plot.title = element_text(hjust = 0.5), plot.subtitle = element_text(hjust = 0.5),)

以月为单位进行分析

df %>%
  count(month) %>%
  ggplot(aes(x = month, y = n, group = 1)) +
  ylim(0, 4000) +
  geom_line(size = 1, color = "darkred") +
  geom_point(size = 2, color = "midnightblue") +
  geom_text(color = "midnightblue", aes(label = n, y = n + 100)) +
  theme_classic() +
  labs(title = "Tracking injury events over time", subtitle = "Note: 2010 & 2020 not a full year", x = "Calendar Year", y = "Number events") +
  theme(panel.grid.major.y = element_blank(), panel.background = element_rect(fill = "whitesmoke"), plot.title = element_text(hjust = 0.5), plot.subtitle = element_text(hjust = 0.5))

注意：季后赛通常是四月中旬，所以一些轮休的球员，一些小伤病的球员通常会选择在季后赛前复出，所以伤病情况会有一个显著的下降，然后是休赛期，到了九月联赛开打，不可避免的伤病会逐渐增多