Exploring values of home runs

homerunvalue.R

# load in two packages

library(tidyverse)
library(WinProbability)

# assume dataset all2018.csv is in current working
# directory -- these functions compute the runs
# expectancies and WPA values

d2018 <- compute.runs.expectancy(2018)
d2018 <- compute.win.probs(d2018)

# focus on home runs

hr <- filter(d2018, EVENT_CD == 23)

# graph of runs values of home runs

ggplot(hr, aes(RUNS.VALUE)) +
  geom_bar(width = 0.05)

# for all players
# find hr count and mean runs values
# graph runs values against HR

hr %>%
  group_by(BAT_ID) %>%
  summarize(N = n(), R = sum(RUNS.VALUE)) %>%
  ggplot(aes(N, R / N)) +
  geom_point()

# for all players
# find hr count and mean WPA value
# graph runs values against HR

hr %>%
  group_by(BAT_ID) %>%
  summarize(N = n(), WPA = sum(abs(WPA))) %>%
  ggplot(aes(N, WPA / N)) +
  geom_point()

# store summaries in a data frame S

hr %>%
  group_by(BAT_ID) %>%
  summarize(N = n(), WPA = sum(abs(WPA))) -> S

# add first and last names from Lahman Master file
# to S, storing in data frame S2

library(Lahman)
library(ggrepel)

inner_join(S, select(Master, retroID, nameFirst, nameLast),
        by = c("BAT_ID" = "retroID"))  %>%
  mutate(Name = paste(nameFirst, nameLast))  %>%
  select(Name, N, WPA) -> S2

# here is a theme for changing attributes of plot title

TH <- theme(
  plot.title = element_text(
    colour = "blue",
    size = 18,
    hjust = 0.5,
    vjust = 0.8,
    angle = 0
  )
)

# graph of HR against mean WPA with interesting
# points labeled

  ggplot(S2, aes(N, WPA / N, label = Name)) +
  geom_point() +
  geom_label_repel(data = filter(S2,  WPA / N > 0.4 |
                              N > 40 |
                              N > 14.5 & WPA /N > .2)) +
    xlab("HR") + ylab("AVERAGE WPA") + TH +
    ggtitle("Scatterplot of HR and Average WPA for 2008 Players")

# focus on comparing two players
# first get Retro ids for both players
# create new data frame hr2 and define labels for
# new factor variable

jd <- "martj006"
kd <- "davik003"

hr2 <- filter(hr, BAT_ID == jd | BAT_ID == kd)
hr2$BAT_ID <- factor(hr2$BAT_ID,
              labels = c("Khris Davis",
                         "J.D. Martinez"))

# parallel dotplot of abs(WPA) for two players

hr2 %>%
  ggplot(aes(abs(WPA))) +
  geom_dotplot() +
  facet_wrap(~ BAT_ID, ncol = 1) +
  theme(text = element_text(size = 16)) +
  ggtitle("Values of Home Runs for Two Sluggers") +
  TH

# parallel dotplot of Runs Value for two players

filter(hr, BAT_ID == jd | BAT_ID == kd) %>%
  ggplot(aes(RUNS.VALUE)) +
  geom_dotplot(dotsize = 0.6) +
  facet_wrap(~ BAT_ID, ncol = 1)

# display average Runs Value and average abs(WPA) for
# two players

hr2 %>%
  group_by(BAT_ID) %>%
  summarize(N = n(),
            Mean_Runs = mean(RUNS.VALUE),
            Mean_WPA = mean(abs(WPA)))

# out of curiousity do a similar thing for 2017
# season (need to have the file all2017.csv in current working directory)

d2017 <- compute.runs.expectancy(2017)
d2017 <- compute.win.probs(d2017)
hr2 <- filter(d2017, EVENT_CD == 23,
              BAT_ID == jd | BAT_ID == kd)
hr2$BAT_ID <- factor(hr2$BAT_ID,
                     labels = c("Khris Davis",
                                "J.D. Martinez"))
hr2 %>%
  group_by(BAT_ID) %>%
  summarize(N = n(),
            Mean_Runs = mean(RUNS.VALUE),
            Mean_WPA = mean(abs(WPA)))

filter(d2017, EVENT_CD == 23) %>%
  summarize(N = n(),
            Mean_Runs = mean(RUNS.VALUE),
            Mean_WPA = mean(abs(WPA)))