player rankings data creation script

positional_adp_merge.R

options(stringsAsFactors = FALSE)
library(jsonlite)
library(dplyr)




## params
years_adp <- 
   c(2018)

years_metrics <-
   c(2016:2017)

fp <-
   '/Users/zenomuscarella/Documents/FF_projs/data/ffstatistics/'

## merging all FFStats player data
temp <- list()

temp[['qb']] <-
   read.csv(paste0(fp, 'qbdata.csv'))

temp[['qb']]$position <-
   'QB'

temp[['rb']] <-
   read.csv(paste0(fp, 'rbdata.csv'))

temp[['rb']]$position <-
   'RB'

temp[['wr']] <-
   read.csv(paste0(fp, 'wrdata.csv'))

temp[['wr']]$position <-
   'WR'

temp[['te']] <-
   read.csv(paste0(fp, 'tedata.csv'))

temp[['te']]$position <-
   'TE'

all_cols <-
   unique(unlist(lapply(temp, names)))


for(i in names(temp)){
   # i <- names(temp)[1]
   names_to_add <- 
      setdiff(all_cols, names(temp[[i]]))
   
   temp[[i]][, names_to_add] <- 
      NA
}


ffstats <- 
   data.frame(do.call(rbind, temp))


## format some of the data 

names(ffstats) <-
   unlist(lapply(tolower(names(ffstats)), 
                 function(x) {
                    y <-
                       gsub('..', '.', x, fixed=TRUE)
                    if(substr(y,nchar(y), nchar(y)) == '.') {
                       return(substr(y,1, nchar(y)-1))
                    } else {
                       return(y)
                       
                    }
                 }
   ))


## check rows match
nrow(ffstats)
sum(unlist(lapply(temp, nrow)))


## subset down to yeras
ffstats <-
   data.frame(ffstats[ffstats$year %in% years_metrics, ])


ffstats$player <- 
   gsub('.', '', ffstats$player, fixed = TRUE)


ffstats$ppg.ppr <-
   ifelse(is.na(ffstats$ppg.ppr), ffstats$ppg.4pt.td, ffstats$ppg.ppr)

ffstats$posrank.ppr <- 
   ifelse(is.na(ffstats$posrank.ppr), ffstats$posrank.4pt.td, ffstats$posrank.ppr)


ffstats <-
   ffstats %>%
   group_by(position) %>%
   mutate(ppg_pos_rank = order(order(ppg.ppr, decreasing=TRUE)))


ffstats <-
   data.frame(ffstats)

#### ffstats data merged! -- could also replace the above w/ the db file but... Zeno doesn't have it 
#### 

## Pull the FFcalc ADP!! 
## FFCalc acutally has a solid API we can use.. no need for importing pesky csv's!

ffcalc <- 
   NULL

for(i in years_adp){
   
   # i <- years[1]
   url <- 
      paste0('https://fantasyfootballcalculator.com/api/v1/adp/ppr?teams=12&year=', i)
   
   temp <- 
      fromJSON(url)
   
   # names(temp) ## highly recommend checking this out -- APIs are cool and this one gives you 
   ## some #good info about what you're pulling
   
   temp <- 
      temp$players
   
   names(temp) <-
      tolower(names(temp))
   
   temp$year <-
      i
  
   temp <-
      temp %>%
      group_by(position) %>%
      mutate(pos_redraft_adp = order(order(adp, decreasing=FALSE)),
             redraft_adp = adp) ## use order order to get psotional adp ranks
   
   
   ffcalc <- rbind(data.frame(temp), ffcalc)
}

ffcalc$player <-
   gsub('.', '', ffcalc$name, fixed =TRUE)


## try to run a dirty merge 

res <- 
   merge(ffstats, ffcalc, by = c('player', 'position'), all.x=TRUE)

res$year_metric <- 
   res$year.x


res$year.x <- 
   NULL

res$year.y <- 
   NULL

res$team <-
   res$team.x

### pull in dlf data

dlf <-
   read.csv(paste0(fp, 'dlfadp players.csv'))

names(dlf) <-
   tolower(names(dlf))


dlf$player <- 
   do.call(rbind, strsplit(dlf$player, ', ', fixed=FALSE))[,1]

dlf$player <- 
   gsub('.', '', dlf$player, fixed=TRUE)

dlf$pos_dyno_adp <- 
   dlf$posadp

dlf$dyno_adp <-
   dlf$adp

res <- 
   merge(res, dlf, by = c('player', 'position'), all.x = TRUE)

res$age <-
   res$age.y   

to_keep <-
   c('player', 'position', 'team','age','year_metric', 
     'pos_redraft_adp', 'pos_dyno_adp','redraft_adp',
     'dyno_adp', 'ppg.ppr', 'ppg_pos_rank')


temp <- 
   res %>%
   select(to_keep) %>%
   group_by(position, year_metric) %>%
   mutate(
      pos_redraft_adp_zscore = scale(pos_redraft_adp, center=TRUE, scale=TRUE),
      pos_dyno_adp_zscore = scale(pos_dyno_adp, center=TRUE, scale=TRUE),
      ppg.ppr_zscore = scale(ppg.ppr, center=TRUE, scale=TRUE),
      ppg_pos_rank_zscore = scale(ppg_pos_rank, center=TRUE, scale=TRUE)
      )
   




write.csv(res[, to_keep], paste0(fp, 'ffstats_playerrankings_data.csv'))