'21 nets performance in games missed by stars r markdown

2021-06-17-nba-bbr.Rmd

---
title: "'21 Brooklyn Nets performance in games missed by stars"
layout: post
categories:
  - notebook
tags:
  - nba, sports
output: 
   html_document
#  md_document:
#    variant: markdown_github+backtick_code_blocks
#    preserve_yaml: true
#    toc: false
#    fig_retina: 2
  
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```



```{r include=FALSE}
library(rvest)
library(cowplot)
library(tidyverse)
library(xml2)
```


```{r}
roster <- "https://www.basketball-reference.com/teams/BRK/2021.html"

html <- read_html(roster)
tables <- html %>% html_nodes("table") %>% html_table()
df <- tables[1] %>% data.frame()

players <- html %>% html_nodes("table") %>% xml_find_all(".//a") %>% xml_attrs()
gamelogs <- unlist(players)
pgl <- gamelogs[grep("gamelog", gamelogs)]  %>% unique() %>% list() 
pgl <- pgl[[1]]

playoffs_pl <- pgl[c(1,2,3,4,6,9,10,14,12,15,18)]
```

```{r }

irving <- "https://www.basketball-reference.com/players/i/irvinky01/gamelog/2021"
harden <- "https://www.basketball-reference.com/players/h/hardeja01/gamelog/2021"
durant <- "https://www.basketball-reference.com/players/d/duranke01/gamelog/2021"
 
get_player_url <- function(path) {
  paste("https://www.basketball-reference.com/",path,sep="")
}

player_urls <- unlist(lapply(FUN=get_player_url, playoffs_pl))

get_data <- function(player_url) {
  html <- read_html(player_url)
  tables <- html %>% html_nodes("table") %>% html_table()
  df <- tables[8] %>% data.frame()
  return(df)
}

```

```{r}

for (p in player_urls) {
   pname <- str_split(p, "/")[[1]][7]
   data <- get_data(p)
   assign(pname, data) 
}
```

```{r}
plist <- list()
for (p in player_urls) {
   pname <- str_split(p, "/")[[1]][7]
   plist <- c(plist, pname)   # Add element to list
}
unlist(plist)

df1 <- get_data(irving)
df2 <- get_data(harden)
df3 <- get_data(durant)

```

```{r}
nets_games_url <- 'https://www.basketball-reference.com/teams/BRK/2021_games.html'
html <- read_html(nets_games_url)
tables <- html %>% html_nodes("table") %>% html_table()

nets_regular_season <- tables[[1]] %>% data.frame()
nets_playoffs <- tables[[2]] %>% data.frame()

games <- nets_regular_season %>% select(G, Date, Var.6, Opponent, Var.8, Tm, Opp)
names(games) <- c("G", "Date", "Home", "Opponent", "WinLoss", "Team", "Opp")
```

```{r}

games$date <- apply(games, 1, FUN = function(x) {substr(x[c("Date")], 6, 20) })
games$date <- as.Date(games$date, c("%B %d, %Y"))

glimpse(games)

games$Team <- as.integer(games$Team)
games$Opp <- as.integer(games$Opp)

games$diff <- games$Team - games$Opp
```

```{r warning=FALSE}
hist(games$diff)
games$home <- games$Home != '@'
table(games$home)
#qplot(games$date, games$diff)
games %>% ggplot(.) + geom_point(aes(y=diff, x=date)) + facet_grid(home ~ .) + geom_smooth(aes(x=date, y=diff))
```

```{r}
irving <- df1 %>% select(Rk, G)
names(irving) <- c("Rk", "Irving")

harden <- df2 %>% select(Rk, G)
names(harden) <- c("Rk", "Harden")

durant <- df3 %>% select(Rk, G)
names(durant) <- c("Rk", "Durant")
```


```{r message=FALSE, warning=FALSE}
m <- irving %>% left_join(harden, by=c("Rk"="Rk"))
m2 <- m %>% left_join(durant, by=c("Rk"="Rk"))
data <- m2 %>% filter(Rk != "Rk")
nrow(data)

data$Irving <- ifelse(is.na(data$Irving), "", data$Irving) 
data$Harden <- ifelse(is.na(data$Harden), "", data$Harden) 
data$Durant <- ifelse(is.na(data$Durant), "", data$Durant) 


data$Irving <- data$Irving != ""
data$Harden <- data$Harden != ""
data$Durant <- data$Durant != ""

data$i <- ifelse(data$Irving, "I", "") 
data$h <- ifelse(data$Harden, "H", "") 
data$d <- ifelse(data$Durant, "D", "") 

data <- data %>% mutate(N_stars = Irving + Harden + Durant)

data <- data %>% mutate(star_combo = paste(i, h, d, sep=""))

dfm <- games %>% left_join(data, by=c("G"="Rk"))

dfm %>% ggplot(.) + geom_point(aes(y=diff, x=date)) + facet_wrap(. ~ N_stars) + geom_smooth(aes(x=date, y=diff)) + theme_cowplot()

dfm %>% group_by(N_stars) %>% summarise(mean(diff), sd(diff), length(diff))

dfm %>% group_by(star_combo) %>% summarise(mean(diff), sd(diff), length(diff))

dfm %>% ggplot(.) + geom_point(aes(y=diff, x=date)) + facet_wrap(. ~ star_combo) + geom_smooth(aes(x=date, y=diff)) + theme_cowplot()
```

```{r warning=FALSE}

dfm %>% ggplot(.) + geom_histogram(aes(x=diff, fill=star_combo), binwidth = 5) + facet_wrap(. ~ star_combo) + theme_cowplot() + labs(x="margin", y="#games")

```

```{r warning=FALSE}

dfm %>% ggplot(.) + geom_density(aes(x=diff, fill=star_combo), binwidth = 5) + facet_wrap(. ~ star_combo) + theme_cowplot() + labs(x="margin", y="#games", fill="Star Combo")

```