My R Experience : Data Analysis of Motorcycle Crash Data in Massachusetts

Mission Statement

Embed to the left of this description is a pdf of "NFL Play by Play Analysis, Linear, Logarithmic, and Random Forest Models", created and presented by myself and a peer in informatics, Brennan Frost. The mission statement of this paper was to investigate the connection between quarterback formations, and outcomes of plays (For example, are certian formations more likely to result in touchdowns or yards gained) using data gathered from NFL Savant which has collected play by play information for every single play in the NFL for more than a decade.

Results Overview

Using linear regression, Logarithmic regression and random forest tree models, we were able to determine these results. For yards gained, using linear regression, running behind the right end, left end and left -tackle were the three most successful rush directions in terms of yards gained, in that order in the 2023 season. For most likely formations to score a touchdown, using logarithmic regression, we were able to determine the order of most likely to least likely to score a touchdown was No huddle, Shotgun, No huddle Shotgun, and Under Center. From our random forest model, we were able to correctly predict 99.15% of plays that would be touchdown with a 88.8% correctly predicted no touchdown rate!