R/DL_FootballDataCoUk_season.R

In tristanbains/rFootballAnalysis:

#' DL_FootballDataCoUk_season
#'
#' download one season for one league from www.football-data.co.uk
#'
#' @export
#' @import tidyverse
#' @param div e.g. "ENG1"
#' @param url e.g. "http://www.football-data.co.uk/mmz4281/1516/E0.csv"
#' @return tibble for one season and one league with all matches, results and bookmaker odds
#'

DL_FootballDataCoUk_season = function(div,url){

  # Get the data:
  Tttt=read.csv(url,stringsAsFactors=FALSE)

  Ttt = Tttt %>%
    filter(!is.na(HomeTeam),
           HomeTeam!="",
           !is.na(AwayTeam)) %>%
    mutate(Div=div,
           Res=sign(FTHG-FTAG),
           Season="",
           Date=ifelse(nchar(Date)==8,as.Date(Date,format="%d/%m/%y"),as.Date(Date,format="%d/%m/%Y")),
           HomeTeam=sub("(^ *)(.*)","\\2",HomeTeam),
           AwayTeam=sub("(^ *)(.*)","\\2",AwayTeam)) %>%
    mutate(Date=as.Date(Date,origin="1970-01-01"))

  # Determine bookmaker columns:
  bcols = tibble(Cols = colnames(Ttt)) %>%
    mutate(Cols2 = substr(Cols,1,(nchar(Cols)-1))) %>%
    group_by(Cols2) %>%
    mutate(Count=n()) %>%
    ungroup() %>%
    filter(Count==3)
  bcols = bcols$Cols
  bcols = c(bcols[grepl("^[[:upper:]]+$", bcols)],bcols[grepl("365", bcols)])

  # Determine columns that contain goals:
  gcols=intersect(colnames(Ttt),c("FTHG","FTAG","Res","HTHG","HTAG"))

  if(length(bcols)>0){
    Tt = Ttt %>%
      dplyr::select_(.dots = c("Div","Season","Date","HomeTeam","AwayTeam",gcols,bcols)) %>%
      gather_("Bookmaker","Odds",bcols) %>%
      mutate(Location = substr(Bookmaker,nchar(Bookmaker),nchar(Bookmaker)),
             Bookmaker = substr(Bookmaker,1,(nchar(Bookmaker)-1))) %>%
      spread(Location,Odds) %>%
      mutate(Overrun = 1/H + 1/D + 1/A) %>%
      mutate(H = ifelse(Overrun<0.95,0,H),
             D = ifelse(Overrun<0.95,0,D),
             A = ifelse(Overrun<0.95,0,A)) %>%
      dplyr::select(-Overrun) %>%
      gather_("Location","Odds",c("A","D","H")) %>%
      group_by(Date,HomeTeam,AwayTeam,Location) %>%
      arrange(-Odds) %>%
      filter(row_number()==1) %>%
      ungroup() %>%
      mutate(Bookmaker_Odds = paste(Bookmaker,Odds,sep="_")) %>%
      dplyr::select(-Bookmaker,-Odds) %>%
      spread(Location,Bookmaker_Odds) %>%
      separate(H,c("BBHO","BHO"),sep="_") %>%
      separate(D,c("BBDO","BDO"),sep="_") %>%
      separate(A,c("BBAO","BAO"),sep="_") %>%
      mutate(BHO=as.numeric(BHO)) %>%
      mutate(BDO=as.numeric(BDO)) %>%
      mutate(BAO=as.numeric(BAO)) %>%
      dplyr::select_(.dots=c("Div","Season","Date","HomeTeam","AwayTeam",gcols,"BHO","BDO","BAO","BBHO","BBDO","BBAO"))
  } else {
    Tt = Ttt %>%
      dplyr::select_(.dots = c("Div","Season","Date","HomeTeam","AwayTeam",gcols))

  }

  Tt = Tt %>%
    mutate(HomeTeam=sub("^ *","",HomeTeam),
           HomeTeam=sub(" *$","",HomeTeam),
           AwayTeam=sub("^ *","",AwayTeam),
           AwayTeam=sub(" *$","",AwayTeam)) %>%
    arrange(Div,Date,HomeTeam)


  return(Tt)

#
#   # determine best odds and bookmakers:
#   cn = colnames(Ttt)
#   bcols = as.data.frame(table(substr(cn,1,nchar(cn)-1)),stringsAsFactors=FALSE) %>%
#     mutate(Var2=substr(Var1,1,3)) %>%
#     group_by(Var2) %>%
#     mutate(n=n()) %>%
#     ungroup() %>%
#     filter(Freq==3,n==1) %>%
#     dplyr::select(Var1)
#   if(nrow(bcols)>0){
#     hcols=intersect(cn,paste(bcols$Var1,"H",sep=""))
#     dcols=intersect(cn,paste(bcols$Var1,"D",sep=""))
#     acols=intersect(cn,paste(bcols$Var1,"A",sep=""))
#     HO = Ttt %>%
#       dplyr::select(one_of(hcols)) %>%
#       mutate(None=0,
#              None2=0)
#     BHO=apply(HO,1,max,na.rm=TRUE)
#     DO = Ttt %>%
#       dplyr::select(one_of(dcols)) %>%
#       mutate(None=0,
#              None2=0)
#     BDO=apply(DO,1,max,na.rm=TRUE)
#     AO = Ttt %>%
#       dplyr::select(one_of(acols)) %>%
#       mutate(None=0,
#              None2=0)
#     #AO.ranked=data.frame(t(apply(-AO, 1, rank, ties.method='min')))
#     #AO.ranked=data.frame(t(apply(-AO, 1, rank, ties.method='first')))
#     BAO=apply(AO,1,max,na.rm=TRUE)
#     OV=1/BHO+1/BDO+1/BAO
#     boolean.OVcorrection=(OV<0.99)
#     BHO2=apply(HO,1,function(x) sort(x, decreasing = TRUE)[2])
#     BDO2=apply(DO,1,function(x) sort(x, decreasing = TRUE)[2])
#     BAO2=apply(AO,1,function(x) sort(x, decreasing = TRUE)[2])
#     # TO DO: correctie om NA en -Inf
#     BBHO=colnames(HO)[apply(HO,1,which.max)]
#     BBDO=colnames(DO)[apply(DO,1,which.max)]
#     BBAO=colnames(AO)[apply(AO,1,which.max)]
#     maxn = function(n) function(x) order(x, decreasing = TRUE)[n]
#     BBHO2=colnames(HO)[apply(HO,1,maxn(2))]
#     BBDO2=colnames(DO)[apply(DO,1,maxn(2))]
#     BBAO2=colnames(AO)[apply(AO,1,maxn(2))]
#     Ttt = Ttt %>%
#       mutate(BHO=ifelse(OV<0.99,BHO2,BHO),
#              BDO=ifelse(OV<0.99,BDO2,BDO),
#              BAO=ifelse(OV<0.99,BAO2,BAO),
#              BHO=ifelse(BHO==0,NA,BHO),
#              BDO=ifelse(BDO==0,NA,BDO),
#              BAO=ifelse(BAO==0,NA,BAO),
#              BHO=round(BHO,2),
#              BDO=round(BDO,2),
#              BAO=round(BAO,2),
#              BBHO=ifelse(OV<0.99,BBHO2,BBHO),
#              BBDO=ifelse(OV<0.99,BBDO2,BBDO),
#              BBAO=ifelse(OV<0.99,BBAO2,BBAO),
#              BBHO=substr(BBHO,1,nchar(BBHO)-1),
#              BBDO=substr(BBDO,1,nchar(BBDO)-1),
#              BBAO=substr(BBAO,1,nchar(BBAO)-1),
#              BBHO=gsub("Non.*",NA,BBHO),
#              BBDO=gsub("Non.*",NA,BBDO),
#              BBAO=gsub("Non.*",NA,BBAO))
#   } else {
#     Ttt = Ttt %>%
#       mutate(BHO=NA,
#              BDO=NA,
#              BAO=NA,
#              BBHO=NA,
#              BBDO=NA,
#              BBAO=NA)
#   }
#   # create final output:
#   cols=c("Div","Season","Date","HomeTeam","AwayTeam","FTHG","FTAG","Res","HTHG","HTAG","BHO","BDO","BAO","BBHO","BBDO","BBAO")
#   cn=intersect(cols,colnames(Ttt))
#   Ttt=Ttt[,cn] %>%
#     mutate(HomeTeam=sub("^ *","",HomeTeam),
#            HomeTeam=sub(" *$","",HomeTeam),
#            AwayTeam=sub("^ *","",AwayTeam),
#            AwayTeam=sub(" *$","",AwayTeam))
#   return(Ttt)
}