bayesball

# loads in the Retrosheet data
load("~/OneDriveBusiness/Retrosheet/pbp.2014.Rdata")

# removes all non-pitches from PITCH_SEQ_TX
pbp.14$pseq <- gsub("[.>123N+*]", "", pbp.14$PITCH_SEQ_TX)

# create a b and s sequence
pbp.14$pseq <-gsub("[BIPV]", "b", pbp.14$pseq)
pbp.14$pseq <-gsub("[CFKLMOQRST]", "s", pbp.14$pseq)

bayesball

groundball.plot <- function(pbp, season){
   require(dplyr)
   require(ggplot2)
   require(car)
   inplay <- filter(pbp, BATTEDBALL_CD == "F" |
                   BATTEDBALL_CD == "G" |
                   BATTEDBALL_CD == "L" |
                   BATTEDBALL_CD == "P")

    inplay <- mutate(inplay,

bayesball

# Script to Learn About Ryan Howard's Batting Ability from a "0 for 35" Slump

# Uses a function from the BayesTestStreak package
# install_github("bayesball/BayesTestStreak")

library(MASS)
library(BayesTestStreak)


# Simulate 500 at-bats with a constant hitting probability p = 0.250 

bayesball

# Daniel Murphy Exercise

# Part I -- learning about Murphy's home run ability
# and updating this knowledge after the NLDS and NLCS

library(ggplot2)
library(LearnBayes)

# career home run data for Murphy 

bayesball

#######################################################################################
# R functions for Paper "Improved Component Predictions of Batting and Pitching Measures"
# Journal of Quantitative Analysis of Sports (2016)
# Jim Albert, albert@bgsu.edu
# functions fit_component_average, plot_avg_results
#           fit_component_obp, plot_obp_results
#           fit_component_fip, plot_fip_results
# require installation of packages Lahman, dplyr, ggplot2, and LearnBayes
########################################################################################

bayesball

# requires packages
# dplyr, Lahman, ggplot2

# some preliminary work

library(dplyr)
library(Lahman)

get.birthyear <- function(player.id){

bayesball

###############################################
# working with complete set of 2015 play-by-play data
# collected using the getData function in the openWAR
# package (retrieves MLBAM GameDay files)
# currently have this saved as a Rdata file
###############################################
load("alldata2015.Rdata")

# computes the run values of all plate appearances

bayesball

library(pitchRx)
library(dplyr)
library(ggplot2)

dat <- scrape(start = "2015-09-05", end = "2015-09-11")
pitches <- inner_join(select(dat$atbat, 
                   batter_name, pitcher_name, inning, 
                   gameday_link, num, url),
                select(dat$pitch,
                   start_speed, pitch_type, sv_id, num, url),

bayesball

fit.model <- function(data){
   # data is a list with two components 
   # - y: binomial counts
   # - n: binomial sample sizes
    require(LearnBayes)
    mode <- laplace(betabinexch, c(1, 1), 
                    cbind(data$y, data$n))$mode
    eta <- exp(mode[1]) / (1 + exp(mode[1]))
    K <- exp(mode[2])
    list(eta=eta, K=K, 

bayesball

# R script for
# "Revisiting Efron and Morris's Data" 
# blog post of February 15, 2016

# in current working directory need a download folder with two subfolders 
# zipped and unzipped
# (For a Windows computer, you need to have the Chadwick cwevent.exe 
# inside the “upzipped” folder.)

library(devtools)