In bayesball/WinProbability: Win Probabilities for Retrosheet

knitr::opts_chunk$set(echo = TRUE,
                      warning = FALSE,
                      message = FALSE)

Introduction

Here is a brief illustration of using the WinProbability package just created on my Github site.

Install the package from Github:

library(devtools)
install_github("bayesball/WinProbability")

Load the package:

library(WinProbability)

Setup

I'm in a folder where I have a raw Retrosheet play-by-play datafile. Note that the file must be of the form allYEAR.csv where YEAR is the value input in the function.

dir()

Compute runs expectancies

The compute.runs.expectancy function:

• Reads in the raw dataset
• Adds a header to the Retrosheet datafile
• Computes the run expectancies and adds these to the datafile

d2018 <- compute.runs.expectancy(2018)

Compute win probabilities

The compute.win.probs function will

• Compute the win probabilities for all plays.
• The input is the Retrosheet play-by-play data frame with run expectancy values included

d2018 <- compute.win.probs(d2018)

Graph the win probabilities for a game

The graph.game function will plot the win probabilities for all plays in a specific game.

plays <- graph.game(d2018, "SLN201805190")

bayesball/WinProbability documentation built on May 9, 2019, 8:42 a.m.