updated to run predictive analytic

Author: ct-analytics

compile fangraphs data.R

@@ -9,12 +9,26 @@ p <- f[str_detect(f,"pitchers")]
 
 b.df <- lapply(b,function(x) {
   fg <- read.csv(x)
+  return(fg)
 })
 
 p.df <- lapply(p,function(x) {
   fg <- read.csv(x)
+  return(fg)
 })
 
 library(data.table)
 batters <- rbindlist(b.df)
 pitchers <- rbindlist(p.df)
+
+pitchers <- pitchers %>%
+  mutate(Date=as.Date(Date)) %>%
+  select(-X) %>%
+  as.data.frame()
+
+batters <- batters %>%
+  mutate(Date=as.Date(Date)) %>%
+  select(-X) %>%
+  as.data.frame()
+
+rm(b.df,p.df,b,p,f)

predictive metrics - pitchers.R

@@ -1,95 +1,43 @@
-library(rvest)
-library(dplyr)
-library(stringr)
-
-month <- "5"
-day <- str_pad(as.character(1:30),2,pad="0")
-
-for (d in day) {
-  date <- as.Date(paste(month,d,"2015",sep="-"),"%m-%d-%Y")
-  date.rotoguru <- paste(month,d,sep="")
-  url <- paste("http://rotoguru1.com/cgi-bin/byday.pl?date=",date.rotoguru,"&game=dd",sep="")
-  rotoguru <- html(url)
-  
-  dk <- rotoguru %>%
-    html_nodes("table") %>%
-    .[[5]] %>%
-    html_table(fill=TRUE) %>%
-    mutate(Player = str_replace(X2,"\\^\\d?",""),
-           Salary = as.numeric(str_replace_all(X4,"\\$|,","")),
-           Points = as.numeric(X3),
-           Date = date) %>%
-    select(Player,Date,Salary,Points) %>%
-    filter(!(is.na(Salary)))
-  
-  if (d==day[1]) {
-    dk.pitchers <- dk
-  } else {
-    dk.pitchers <- rbind(dk.pitchers,dk)
-  }
-}
+setwd("/Users/christopherteixeira/Documents/draftkings optimization/")
+source("salary data till now.R")
+source("compile fangraphs data.R")
 
-rm(dk,rotoguru,url,date,date.rotoguru)
-stats <- list(pitching="pit",batting="bat")
-
-day <- day[!day %in% c("06","23")]
-
-for (d in day) {
-  date <- as.Date(paste(month,d,"2015",sep="-"),"%m-%d-%Y")
-  
-  url <- paste("http://www.fangraphs.com/leaders.aspx?pos=all&stats=",stats$pitching,"&lg=all&qual=0&type=8&season=2015&month=0&season1=2015&ind=0&team=0&rost=0&age=0&filter=&players=p",format(date,"%Y-%m-%d"),"page=1_50",sep="")
-  raw <- html(url)
-  
-  fg.dashboard <- raw %>%
-    html_nodes("table") %>%
-    .[[33]] %>%
-    html_table(fill=TRUE) %>%
-    select(-1) %>%
-    mutate(Date = date)
-  
-  url <- paste("http://www.fangraphs.com/leaders.aspx?pos=all&stats=",stats$pitching,"&lg=all&qual=0&type=3&season=2015&month=0&season1=2015&ind=0&team=0&rost=0&age=0&filter=&players=p",format(date,"%Y-%m-%d"),"page=1_50",sep="")
-  raw <- html(url)
-  
-  fg.winprobability <- raw %>%
-    html_nodes("table") %>%
-    .[[33]] %>%
-    html_table(fill=TRUE) %>%
-    select(-1) %>%
-    mutate(Date = date)
-  
-  if (d==day[1]) {
-    fangraphs.dashboard <- fg.dashboard
-    fangraphs.winprobability <- fg.winprobability
-  } else {
-    fangraphs.dashboard <- rbind(fangraphs.dashboard,fg.dashboard)
-    fangraphs.winprobability <- rbind(fangraphs.winprobability,fg.winprobability)
-  }
-}
+library(dplyr)
 
-rm(d,fg.dashboard,fg.winprobability,date,date.rotoguru,raw,url,stats)
+df <- inner_join(draftkings,
+                 pitchers,
+                 by=c("Player"="Name","Date"="Date")) 
 
-pitchers <- inner_join(dk.pitchers,fangraphs.dashboard,by=c("Player"="Name","Date"="Date")) %>%
-  inner_join(fangraphs.winprobability,by=c("Player"="Name","Date"="Date")) %>%
-  rename(Team=Team.x,
-         K.per.9=`K/9`,
-         BB.per.9=`BB/9`,
-         HR.per.9=`HR/9`,
-         WPA.minus=`-WPA`,
-         WPA.plus=`+WPA`,
-         WPA.per.LI=`WPA/LI`) %>%
-  mutate(LOB.pct=as.numeric(str_replace(`LOB%`,'%','')),
-         GB.pct=as.numeric(str_replace(`GB%`,'%','')),
-         HR.per.FB=as.numeric(str_replace(`HR/FB`,'%',''))) %>%
-  select(-c(Team.y,`LOB%`,`GB%`,`HR/FB`))
+numeric.vars <- names(df)[sapply(df,is.numeric)]
+character.vars <- names(df)[!sapply(df,is.numeric)]
 
 library(ggplot2)
-ggplot(data=pitchers, aes(y=Points)) +
-  geom_point(aes(x=Game.Score))
+g <- ggplot(data=df, aes(y=Points)) 
 
-ggplot(data=pitchers, aes(y=Points)) +
-  geom_point(aes(x=WPA))
+g + geom_point(aes(x=WAR))
+g + geom_point(aes(x=xFIP))
+g + geom_point(aes(x=FIP))
 
-correlations <- as.data.frame(t(cor(pitchers$Points,select(pitchers,-Team,-Player,-Points))))
+correlations <- as.data.frame(t(cor(df$Points,select(df,-Team,-Player,-Points,-Date,-Position))))
 names(correlations) <- c("Correlation")
 arrange(correlations,Correlation)
-rownames(correlations)
+correlations$Statistic <- rownames(correlations)
+
+library(DT)
+datatable(correlations %>% select(Correlation))
+
+library(rCharts)
+h <- rPlot(Points ~ WAR,
+      data=df,
+      color="Position",
+      type = "point")
+h$addControls("x",value="WAR",values=numeric.vars)
+h$addControls("color", value = "Position", values = character.vars)
+h
+
+h <- dPlot(Points ~ WAR,
+           groups = 'Position',
+           data=df,
+           type = "bubble")
+h$legend(x = 200,y = 10,width = 500,height = 20,horizontalAlign = "right")
+h

salary data till now.R

@@ -3,7 +3,7 @@ library(dplyr)
 library(stringr)
 library(data.table)
 
-start.date <- as.Date("06-01-2015","%m-%d-%Y")
+start.date <- as.Date("05-01-2015","%m-%d-%Y")
 today <- Sys.Date()
 days.to.retrieve <- seq(start.date,today,"day")
 
@@ -19,10 +19,12 @@ dk.full <- lapply(days.to.retrieve,function(d) {
     mutate(Player = str_replace(X2,"\\^\\d",""),
            Salary = as.numeric(str_replace_all(X4,"\\$|,","")),
            Points = as.numeric(X3),
-           Date = date) %>%
+           Date = d) %>%
     select(Player,Date,Salary,Points) %>%
     filter(!(is.na(Salary)))
   return(dk)
 })
 
-draftkings <- rbindlist(dk.full)
+draftkings <- as.data.frame(rbindlist(dk.full))
+
+rm(dk.full,days.to.retrieve,start.date,today)