unbeatenruns

#### Note - this isn't the cleanest code, but it works fine

#### install packages
devtools::install_github('jalapic/engsoccerdata', username = "jalapic")
library(engsoccerdata)
devtools::install_github('jalapic/curleylab')
library(curleylab) #just for a ggplot theme


head(engsoccerdata2)

df<-
  rbind(
  engsoccerdata2 %>% filter(Season!=1939) %>% mutate(venue='home') %>% select(Date,Season,team=home, opp=visitor, FT, venue, gd=goaldif),
  engsoccerdata2 %>% filter(Season!=1939) %>% mutate(goaldif1=-goaldif, venue='away') %>% select(Date,Season,team=visitor, FT, venue, opp=home, gd=goaldif1)
)
  


head(df)
df$Date <- as.Date(df$Date, format="%Y-%m-%d")

# Functions to  get sequences

fun1<-function(tmp){
ddf <- data.frame(winlengths = rle(tmp$win)$lengths, winvalues = rle(tmp$win)$values)
ddf.expanded <- ddf[rep(row.names(ddf), ddf$winlengths), 1:2]
tmp$winlengths <- ddf.expanded[,1]
tmp$winvalues <- ddf.expanded[,2]
return(tmp)
}

fun2<-function(tmp){
ddf<-data.frame(winlengths = rle(tmp$win)$lengths, winvalues = rle(tmp$win)$values)
ddf<-ddf %>% filter(winvalues==T) %>% arrange(desc(winlengths))
return(ddf)
}

###############################


#### Win Sequences

df1 <-df %>%
  mutate(win = ifelse(gd>0, T, F))%>% 
  group_by(team) %>%
  arrange(team,Date) 


df1x <- lapply(split(df1, df1$team), fun1)
df1x <- do.call('rbind', df1x)
head(df1x)

df1y <- lapply(split(df1, df1$team), fun2)
df1y <- Map(cbind, df1y, team = names(split(df1,df1$team)))
df1y <- do.call('rbind', df1y)
table(df1y$winlengths)

library(ggplot2)
head(df1y)
ggplot(df1y,aes(x=factor(winlengths))) + geom_bar() + geom_text(stat='bin',aes(label=..count..),vjust=-.7) + 
  curleytheme()  + theme(panel.grid.major.x = element_blank())  +
  xlab("Consecutive Wins") + theme(axis.text.x  = element_text(size=20), axis.text.y  = element_blank(),
                                   axis.title.x  = element_text(size=20)) + ggtitle("Frequency of Winning Runs in English Football") +
  ylab("")



##### Frequency of Unbeaten Runs

df1 <-df %>%
  mutate(win = ifelse(gd>=0, T, F))%>% 
  group_by(team) %>%
  arrange(team,Date) 


df1x <- lapply(split(df1, df1$team), fun1)
df1x <- do.call('rbind', df1x)
head(df1x)

df1y <- lapply(split(df1, df1$team), fun2)
df1y <- Map(cbind, df1y, team = names(split(df1,df1$team)))
df1y <- do.call('rbind', df1y)
table(df1y$winlengths)



library(ggplot2)
head(df1y)

ggplot(df1y,aes(x=factor(winlengths))) + geom_bar() + geom_text(stat='bin',aes(label=..count..),vjust=-.7) + 
  curleytheme()  + theme(panel.grid.major.x = element_blank())  +
  xlab("Consecutive Wins") + theme(axis.text.x  = element_text(size=20), axis.text.y  = element_blank(),
                                   axis.title.x  = element_text(size=20)) + ggtitle("Frequency of Unbeaten Runs in English Football") +
  ylim(0,38000)+  ylab("")