Show simple item record

dc.contributor.authorSmith, Taylor Bradley
dc.date.accessioned2018-09-28T04:30:13Z
dc.date.available2018-09-28T04:30:13Z
dc.date.issued2018-05
dc.identifier.othersmith_taylor_b_201805_ms
dc.identifier.urihttp://purl.galileo.usg.edu/uga_etd/smith_taylor_b_201805_ms
dc.identifier.urihttp://hdl.handle.net/10724/38559
dc.description.abstractI utilize Major League Baseball Statcast data from 2015-2017 to build batted ball classifiers using state-of-the-art gradient boosting trees in conjunction with hyperparameter optimization techniques. Visual and numeric summaries of the model results are used to glean insights into batted balls in MLB. Further, the model framework is used to create new batting and pitching metrics with demonstrated advantages over previously used metrics. Using the batted ball classifiers and the introduced metrics, I investigate the "Juiced Ball" and "Fly Ball Revolution" phenomena in MLB, quantify the respective impacts of both phenomena, and present a manner for evaluating batter and pitcher performance across different ball environments.
dc.languageeng
dc.publisheruga
dc.rightspublic
dc.subjectsabermetrics
dc.subjectjuiced ball
dc.subjectMajor League Baseball
dc.subjecthyperparameter optimization
dc.subjectgradient boosting trees
dc.subjectbaseball statistics
dc.titleAn expected outcome framework for evaluating batting and pitching performance in major league baseball with applications to the "juiced ball" and the "fly ball revolution"
dc.typeThesis
dc.description.degreeMS
dc.description.departmentStatistics
dc.description.majorStatistics
dc.description.advisorJason Anastasopoulos
dc.description.committeeJason Anastasopoulos
dc.description.committeeJaxk Reeves
dc.description.committeeNicole Lazar


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record