FIT5147 Data Exploration Project Raw Code
Nicholas Pennell, 30338913
library(pitchRx) # data pack
library(RSQLite) # loading in data of c drive
library(readr) # read in files
library(dplyr) # cleaing data
library(tidyr) # cleaning data
library(forcats) # cleaning data
library(lubridate) # cleaing dates
library(outliers) # outliers
library(corrplot) # correlation plot
library(RColorBrewer) # colour package
library(viridis) # colour package
library(ggplot2) # plots
library(kableExtra) # html tabel
library(ggthemes) # plot themes
library(tinytex) # knit to pdf
library(knitr) # kniting
library(prettydoc) # rmarkdown themeData Wrangling
The data wrangling process is split into several steps
- Get: Sourcing the data and reading it into in R,
- Understand: Add any new variable, examine and alternating variables, and make sure we know what each variable is.
- Tidy: This is completed with serval packages,
dplyr,tidyr,lubridateandRSQLite.
Get
database.2010 <- src_sqlite("C:/Data/Baseball/pitchRx 2010.sqlite3", create = FALSE)
database.2011 <- src_sqlite("C:/Data/Baseball/pitchRx 2011.sqlite3", create = FALSE)
database.2012 <- src_sqlite("C:/Data/Baseball/pitchRx 2012.sqlite3", create = FALSE)
database.2013 <- src_sqlite("C:/Data/Baseball/pitchRx 2013.sqlite3", create = FALSE)
database.2014 <- src_sqlite("C:/Data/Baseball/pitchRx 2014.sqlite3", create = FALSE)
database.2015 <- src_sqlite("C:/Data/Baseball/pitchRx 2015.sqlite3", create = FALSE)
database.2016 <- src_sqlite("C:/Data/Baseball/pitchRx 2016.sqlite3", create = FALSE)
database.2017 <- src_sqlite("C:/Data/Baseball/pitchRx 2017.sqlite3", create = FALSE)
database.2018 <- src_sqlite("C:/Data/Baseball/pitchRx 2018.sqlite3", create = FALSE)# 2010
# pitch table
pitches.2010 <- tbl(database.2010, "pitch") %>% as.data.frame()
pitches.2010 <- pitches.2010[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2010 <- tbl(database.2010, "atbat") %>% as.data.frame()
atbat.2010 <- atbat.2010[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2010 <- inner_join(pitches.2010, atbat.2010, by = c("num", "gameday_link"))
# filter
chris.sale.2010 <- total.2010 %>% filter(pitcher_name == "Chris Sale")
# 2011
# pitch table
pitches.2011 <- tbl(database.2011, "pitch") %>% as.data.frame()
pitches.2011 <- pitches.2011[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2011 <- tbl(database.2011, "atbat") %>% as.data.frame()
atbat.2011 <- atbat.2011[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2011 <- inner_join(pitches.2011, atbat.2011, by = c("num", "gameday_link"))
# filter
chris.sale.2011 <- total.2011 %>% filter(pitcher_name == "Chris Sale")
# 2012
# pitch table
pitches.2012 <- tbl(database.2012, "pitch") %>% as.data.frame()
pitches.2012 <- pitches.2012[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2012 <- tbl(database.2012, "atbat") %>% as.data.frame()
atbat.2012 <- atbat.2012[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2012 <- inner_join(pitches.2012, atbat.2012, by = c("num", "gameday_link"))
# filter
chris.sale.2012 <- total.2012 %>% filter(pitcher_name == "Chris Sale")
# 2013
# pitch table
pitches.2013 <- tbl(database.2013, "pitch") %>% as.data.frame()
pitches.2013 <- pitches.2013[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2013 <- tbl(database.2013, "atbat") %>% as.data.frame()
atbat.2013 <- atbat.2013[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2013 <- inner_join(pitches.2013, atbat.2013, by = c("num", "gameday_link"))
# filter
chris.sale.2013 <- total.2013 %>% filter(pitcher_name == "Chris Sale")
# 2014
# pitch table
pitches.2014 <- tbl(database.2014, "pitch") %>% as.data.frame()
pitches.2014 <- pitches.2014[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2014 <- tbl(database.2014, "atbat") %>% as.data.frame()
atbat.2014 <- atbat.2014[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2014 <- inner_join(pitches.2014, atbat.2014, by = c("num", "gameday_link"))
# filter
chris.sale.2014 <- total.2014 %>% filter(pitcher_name == "Chris Sale")
# 2015
# pitch table
pitches.2015 <- tbl(database.2015, "pitch") %>% as.data.frame()
pitches.2015 <- pitches.2015[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2015 <- tbl(database.2015, "atbat") %>% as.data.frame()
atbat.2015 <- atbat.2015[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2015 <- inner_join(pitches.2015, atbat.2015, by = c("num", "gameday_link"))
# filter
chris.sale.2015 <- total.2015 %>% filter(pitcher_name == "Chris Sale")
# 2016
# pitch table
pitches.2016 <- tbl(database.2016, "pitch") %>% as.data.frame()
pitches.2016 <- pitches.2016[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2016 <- tbl(database.2016, "atbat") %>% as.data.frame()
atbat.2016 <- atbat.2016[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2016 <- inner_join(pitches.2016, atbat.2016, by = c("num", "gameday_link"))
# filter
chris.sale.2016 <- total.2016 %>% filter(pitcher_name == "Chris Sale")
# 2017
# pitch table
pitches.2017 <- tbl(database.2017, "pitch") %>% as.data.frame()
pitches.2017 <- pitches.2017[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2017 <- tbl(database.2017, "atbat") %>% as.data.frame()
atbat.2017 <- atbat.2017[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2017 <- inner_join(pitches.2017, atbat.2017, by = c("num", "gameday_link"))
# filter
chris.sale.2017 <- total.2017 %>% filter(pitcher_name == "Chris Sale")
# 2018
# pitch table
pitches.2018 <- tbl(database.2018, "pitch") %>% as.data.frame()
pitches.2018 <- pitches.2018[ , c(1, 3, 4, 10, 16, 17, 31, 30, 33 ,42:47)]
# atbat table
atbat.2018 <- tbl(database.2018, "atbat") %>% as.data.frame()
atbat.2018 <- atbat.2018[ , c(1, 2, 3, 6, 9, 11, 12, 14, 16, 17, 19, 20, 24, 25, 27, 28)]
# merge
total.2018 <- inner_join(pitches.2018, atbat.2018, by = c("num", "gameday_link"))
# filter
chris.sale.2018 <- total.2018 %>% filter(pitcher_name == "Chris Sale")# total merge
chris.sale.total <- chris.sale.2010 %>% bind_rows(chris.sale.2011)
chris.sale.total <- chris.sale.total %>% bind_rows(chris.sale.2012)
chris.sale.total <- chris.sale.total %>% bind_rows(chris.sale.2013)
chris.sale.total <- chris.sale.total %>% bind_rows(chris.sale.2014)
chris.sale.total <- chris.sale.total %>% bind_rows(chris.sale.2015)
chris.sale.total <- chris.sale.total %>% bind_rows(chris.sale.2016)
chris.sale.total <- chris.sale.total %>% bind_rows(chris.sale.2017)
chris.sale.total <- chris.sale.total %>% bind_rows(chris.sale.2018)Understand
# pitch Outcome
# strikes
chris.sale.total$pitch_outcome <- ifelse(chris.sale.total$type == "S", chris.sale.total$des, NA)
# Balls
chris.sale.total$pitch_outcome <- ifelse(chris.sale.total$type == "B",
"Ball", chris.sale.total$pitch_outcome)
# balls in play
chris.sale.total$pitch_outcome <- ifelse(chris.sale.total$type == "X",
chris.sale.total$event, chris.sale.total$pitch_outcome)
# Team
chris.sale.total$team <- ifelse(chris.sale.total$date > 2016, "Boston", "White Sox")
chris.sale.total$team <- factor(chris.sale.total$team, levels = c("White Sox", "Boston"))# Changing variable type
# To factor
chris.sale.total[, c("des", "type", "pitch_type", "count", "o", "stand", "p_throws", "event",
"inning_side", "outcome")] <-
lapply(chris.sale.total[, c("des", "type", "pitch_type", "count", "o", "stand", "p_throws",
"event","inning_side", "pitch_outcome")], as.factor)
# To Logical
chris.sale.total[, c("on_1b", "on_2b", "on_3b")] <-
lapply(chris.sale.total[, c("on_1b", "on_2b", "on_3b")], as.logical)
# Replacing NA, with FALSE
chris.sale.total$on_1b <- ifelse(is.na(chris.sale.total$on_1b), FALSE, chris.sale.total$on_1b)
chris.sale.total$on_2b <- ifelse(is.na(chris.sale.total$on_2b), FALSE, chris.sale.total$on_2b)
chris.sale.total$on_3b <- ifelse(is.na(chris.sale.total$on_3b), FALSE, chris.sale.total$on_3b)
# To Date
chris.sale.total[, "date"] <- chris.sale.total[, "date"] %>% ymd()# Home team and away team NA to zero
chris.sale.total$home_team_runs <- ifelse(is.na(chris.sale.total$home_team_runs),
0, chris.sale.total$home_team_runs)
chris.sale.total$away_team_runs <- ifelse(is.na(chris.sale.total$away_team_runs),
0, chris.sale.total$away_team_runs)
# Type
levels(chris.sale.total$type) <- list("Ball" = "B",
"Strike/Fouls" = "S",
"Ball in play" = "X")
# Pitch Type
levels(chris.sale.total$pitch_type) <- list("Two-Seam Fastball" = "FT",
"Slider" = "SL",
"Changeup" = "CH",
"Four-Seam Fastball" = "FF",
"Intentional Walk" = "IN",
"Fastball" = "FA",
"Unknow" = "UN",
"Pitch Out" = "PO",
"Splitter" = "FS",
"Curveball" = "CU")
# Outcome
chris.sale.total$pitch_outcome <- chris.sale.total$pitch_outcome %>%
fct_collapse("Ball" ="Ball",
"Strike Looking" = "Called Strike",
"Swinging Strike" = c("Missed Bunt",
"Swinging Strike",
"Swinging Strike (Blocked)"),
"Foul" = c("Foul", "Foul (Runner Going)",
"Foul Bunt", "Foul Tip"),
"Single" = "Single",
"Double" = "Double",
"Triple" = "Triple",
"Home Run" = "Home Run",
"Out" = c("Lineout", "Grounded Into DP",
"Flyout", "Groundout", "Sac Bunt",
"Forceout", "Pop Out", "Bunt Groundout",
"Bunt Pop Out", "Double Play",
"Sac Fly",
"Fielders Choice Out", "Triple Play",
"Fielders Choice"),
"Error" = c("Field Error", "Batter Interference",
"Fan interference"))
# stand
levels(chris.sale.total$stand) <- list("Left" = "L",
"Right" = "R") varible <- read_csv("varibles.csv")
varible$`Outcome of the pitch` <- ifelse(is.na(varible$X4),
varible$`Outcome of the pitch`,
paste(varible$`Outcome of the pitch`, varible$X4))
varible[, 1:3] %>% kable() %>% kable_styling("striped", full_width = F)| Des | factor | Outcome of the pitch |
|---|---|---|
| Id | numeric | unique identification number per pitch within a game. |
| type | factor | weather the pitch was a ball strike or foul |
| start_speed | numeric | speed when the pitch is thrown in mph |
| px | numeric | horizontal distance in feet from the centre of home plate as the ball crosses home plate |
| pz | numeric | vertical distance in feet from the centre of home plate as the ball crosses home plate |
| type_confidence | numeric | the value of weight to the probability the pitch type is correct |
| pitch_type | factor | most probable pitch type thrown |
| nasty | numeric | how hard a ball is to hit |
| num | numeric | used to help join tables |
| on_1b | logical | if a player is on first base |
| on_2b | logical | if a player is on second base |
| on_3b | logical | if a player is on third base |
| count | factor | how many balls and strikes there are |
| gameday_link | character | link to gameday data (used to join tables) |
| pitcher | numeric | id code for the pitcher |
| batter | numeric | id code for the batter |
| o | factor | number of outs |
| stand | factor | if the batter is left or right handed |
| throws | factor | if the pitcher left or right handed |
| atbat_des | character | what the outcome of the at bat was |
| event | factor | simple outcome |
| home_team_runs | numeric | score for the home team |
| away_team_runs | numeric | score for the away team |
| inning_side | factor | if it’s the top or bottom of an innings |
| inning | numeric | what innings it is |
| batter_name | character | name of the batter |
| pitcher_name | character | name of the pitcher |
| date | date | date |
| outcome | factor | added simpler outcome variable |
| team | factor | what team does Sale play for. |
Tidy
Data Checking
# Creating data frame to plot all numerical baribles
plot.table <- chris.sale.total[, c("nasty", "inning", "type_confidence", "start_speed")] %>%
gather(key = "variable", value = "value")
# plot 1
ggplot(data = plot.table, aes(x = variable, y = value)) + geom_violin() +
geom_boxplot(width=0.1, outlier.colour = 'red') +
facet_wrap( ~ variable, scales = "free") + theme_fivethirtyeight()
Plot 1: Outliers
# Function to Remove Outliers Using Outlier Detection
outlier.qr <- function(data, var_name)
{
data$var_name <- eval(substitute(var_name), data)
iqr <- IQR(data$`var_name`, na.rm = TRUE)
qr <- quantile(data$var_name, c(0.25,0.75))
Q1 <- qr[[1]] - 1.5*iqr
Q3 <- qr[[2]] + 1.5*iqr
data <- data %>% filter(var_name < Q3 & var_name > Q1)
return(data)
}
# removing outliers form type_confidence
chris.sale.total <- outlier.qr(chris.sale.total, type_confidence)# list of numerical varible
numerical.varible <- c("start_speed", "px", "pz", "nasty", "inning")
# colour brewer
pitch_col <- brewer.pal(length(unique(chris.sale.total$pitch_type)),"Set1")
# pair
pairs(chris.sale.total[, numerical.varible], lower.panel = NULL, col = pitch_col)
# add ledgend
par(xpd = TRUE)
legend(x = 0.05, y = 0.5, cex = 1, legend = as.character(levels(chris.sale.total$pitch_type)),
fill = unique(pitch_col))
Plot 2: Corelation Plot 1
par(xpd = NA)# cor plot
chris.sale.total[, numerical.varible] %>% cor() %>% corrplot(method = "color", type="upper")
Plot 3: Corelation PLot 2
Data Exploration
# Formate data
# Pitches by date and type
pitch.by.year <- table(year(chris.sale.total$date), chris.sale.total$pitch_type,
dnn = c("year", "pitch_type")) %>% as.data.frame()
# Pitches by date
total.pitch.year <- table(year(chris.sale.total$date), dnn = "year") %>% as.data.frame()
# Join and order
pitch.by.year <- full_join(pitch.by.year, total.pitch.year, by = "year")
# Formate and adding percentage as well as year days
colnames(pitch.by.year) <- c("year", "pitch_type", "type_freq", "year_freq")
pitch.by.year$perc <- pitch.by.year$type_freq / pitch.by.year$year_freq
pitch.by.year$year <- pitch.by.year$year %>% as.Date("%Y")
pitch.by.year$year <- pitch.by.year$year %>% year()
# Removing zeros
pitch.by.year <- pitch.by.year %>% filter(type_freq != 0) # plot
ggplot(pitch.by.year, aes(x = year, y = perc)) +
geom_line(aes(colour = pitch_type), size = 1) +
geom_point(fill = "black", aes(colour = pitch_type)) +
geom_vline(xintercept = 2016.5, linetype = "dashed", colour = "#666666") +
geom_text(x = 2016.7, y = 0.6, label = "traded", colour = "#666666", size = 6) +
theme_fivethirtyeight() + scale_colour_brewer(palette = 'Set1')
Plot 4: Frequncey By Year
# Formate data
# Pitches by date and type
pitch.by.date <- table(chris.sale.total$date, chris.sale.total$pitch_type,
dnn = c("date", "pitch_type")) %>% as.data.frame()
# Pitches by date
total.pitch.date <- table(chris.sale.total$date, dnn = "date") %>% as.data.frame()
# Join and order
pitch.by.date <- full_join(pitch.by.date, total.pitch.date, by = "date")
pitch.by.date <- pitch.by.date %>% arrange(date)
# Formate and adding percentage as well as year days
colnames(pitch.by.date) <- c("date", "pitch_type", "type_freq", "date_freq")
pitch.by.date$perc <- pitch.by.date$type_freq / pitch.by.date$date_freq
pitch.by.date$date <- pitch.by.date$date %>% ymd()
pitch.by.date$date.yd <- pitch.by.date$date %>% yday()
# Nemoving zeros
pitch.by.date <- pitch.by.date %>% filter(type_freq != 0)
# time series (ggplot)
ggplot(pitch.by.date, aes(x = date.yd, y = perc)) +
geom_line(aes(colour = pitch_type), size = 1) +
theme_grey() + facet_grid(year(date)~pitch_type) +
scale_colour_brewer(palette = 'Set1') + xlab("Year Days") + ylab("Percentage") +
theme_fivethirtyeight()
Plot 5: Frequncey By Year Days
# Home plate and ave. strike zone
# ave strike zone
strike.zone <- geom_rect(aes(xmin = -0.95, xmax = 0.95, ymin = 1.6, ymax = 3.5), fill = NA,
color = "#ff6116", size = 0.4)
# home plate
hp.x <- c(-0.95, -0.95, 0, 0.95, 0.95)
hp.y <- c(0, -0.475, -0.95, -0.475, 0)
hp <- data.frame(hp.x, hp.y)
# Join
home.plate <- geom_polygon(data = hp, aes(x = hp.x, y = hp.y), fill = NA,
color = "#ff6116", size = 0.5) # Hex plot
# All Pitches
ggplot(data = chris.sale.total, aes(x = px, y = pz)) + geom_hex() + coord_fixed() +
strike.zone + home.plate + scale_fill_viridis() + theme_fivethirtyeight()
Plot 6: Location of All Pitches
# facet by pitch type
# Finding tally
pitch.type.tally <- chris.sale.total %>% group_by(pitch_type) %>% tally()
# plot
ggplot(data = chris.sale.total, aes(x = px, y = pz)) + geom_hex() + coord_fixed() +
strike.zone + home.plate + scale_fill_viridis() + facet_grid(~pitch_type) +
theme_fivethirtyeight() +
geom_label(data = pitch.type.tally, aes(label=paste('n=', n)),
x = -4.25, y = 5, vjust = 0, hjust = 0)
Plot 7: Location by Pitch Type
# stand
# stand and pitch type
# Finding tally
pitch.stand.type.tally <- chris.sale.total %>% group_by(pitch_type, stand) %>% tally()
# plot location
ggplot(data = chris.sale.total, aes(x = px, y = pz)) + geom_hex() + coord_fixed() +
strike.zone + home.plate + scale_fill_viridis() + facet_grid(stand~pitch_type) +
theme_fivethirtyeight() +
geom_label(data = pitch.stand.type.tally, aes(label=paste('n=', n)),
x = -4.25, y = 5, vjust = 0, hjust = 0)
Plot 8: Location by Pitch Type and Stand
# formate
stand.freq <- chris.sale.total %>% group_by(stand) %>% tally()
type.stand.freq <- stand.freq %>% full_join(pitch.stand.type.tally, by = "stand")
colnames(type.stand.freq) <- c("stand", "stand.total", "pitch_type", "stand_type_total")
type.stand.freq$perc <- type.stand.freq$stand_type_total / type.stand.freq$stand.total
# pitch type freqency plot
ggplot(data = type.stand.freq, aes(x = pitch_type, y = perc)) + coord_flip() +
geom_bar(aes(fill = pitch_type), stat = "identity") + facet_grid(.~stand) +
theme_fivethirtyeight() + scale_fill_brewer(palette = 'Set1') +
scale_y_continuous(labels = scales::percent)
Plot 9: Freqcency of Pitches by Stand
# team
# stand and pitch type
# Finding tally
pitch.team.type.tally <- chris.sale.total %>% group_by(pitch_type, team) %>% tally()
# plot
ggplot(data = chris.sale.total, aes(x = px, y = pz)) + geom_hex() + coord_fixed() +
strike.zone + home.plate + scale_fill_viridis() + facet_grid(team~pitch_type) +
theme_fivethirtyeight() +
geom_label(data = pitch.team.type.tally, aes(label=paste('n=', n)),
x = -4.25, y = 5, vjust = 0, hjust = 0)
Plot 10: Location by Pitch Type and Team
# formate
holder <- chris.sale.total %>% filter(year(date) == 2016 | year(date) == 2017)
pitch.outcome.tally <- holder %>% group_by(pitch_outcome, year(date)) %>% tally()
# plot
ggplot(pitch.outcome.tally, aes(x = pitch_outcome, y = n)) +
geom_bar(aes(fill = pitch_outcome), stat = "identity") + coord_flip() +
facet_grid(.~ `year(date)`) + scale_fill_brewer(palette = 'Set3') + theme_fivethirtyeight() +
geom_label(data = pitch.outcome.tally, aes(label=n))
Plot 11: Outcome by Year