WARDataManipulation

#1 import libraries

library(tidyverse)
library(ggplot2)
library(baseballr)
library(corrplot)
library(lubridate)
library(IDPmisc)

#Load 3 years FIP data
# FanGraphs WARleaders page, https://www.fangraphs.com/warleaders.aspx?season=2021&team=all

players2021 <- read.csv("C:/Users/~FIPWAR2021.csv")%>%
  rename(Name = ï..Name,WAR2021 = Total.WAR) %>%
  mutate(rank21 = dense_rank(desc(WAR2021))
  ) %>%
  select(Pos,playerid, rank21, WAR2021)

players2020 <- read.csv("C:/Users/~/FIPWAR2020.csv")%>%
  rename(Name = ï..Name,WAR2020 = Total.WAR) %>%
  mutate(rank20 = dense_rank(desc(WAR2020))
  ) %>%
  select(playerid, rank20, WAR2020)

players2019 <- read.csv("C:/Users/~/FIPWAR2019.csv")%>%
  rename(Name = ï..Name,WAR2019 = Total.WAR) %>%
  mutate(rank19 = dense_rank(desc(WAR2019))
  ) %>%
  select(Name, playerid,rank19, WAR2019)

WAROne <- players2019 %>%
  inner_join(players2020, by=("playerid"))

WARFinal <- WAROne %>%
  inner_join(players2021, by=("playerid")) %>%
  mutate(rankchange = rank19 - rank21,
         warchange = (WAR2021 - WAR2019)/WAR2019,
         playerid = as.character(playerid))


# Getting Age info
# https://www.smartfantasybaseball.com/tools/

playerids <- read.csv("C:/Users/danjp/~/SFBB-Player-ID-Map.csv")%>%
  rename(playerid = IDFANGRAPHS) %>%
  mutate(age1 = as.Date(parse_date_time(BIRTHDATE, c('mdy', 'HMS'))),
         age2021 = trunc(as.numeric(difftime("2021-7-19",age1, units = "weeks"))/52.25),
         age2019 = age2021 - 2,
         age2025 = age2021 + 4) %>%
  select(playerid,age2021,age2019)


#Identifying Top 15 newcomers 2025

WARFinal2 <- WARFinal %>%
  inner_join(playerids, by=("playerid")) %>%
  distinct() %>%
  select(Name,Pos,rank21, age2021, rankchange, warchange) %>%
  filter(age2021 < 25, rank21 < 51)