Mini Project Idea 1
+My idea is centered around the current day NFL and the analytics behind “going for it” on fourth down. The project will center around this question: “Do NFL teams go for it on fourth down in the right scenarios. What teams are most aggresive in their play calling?”
+The data is retrievable because NFL play-by-play data is available publicly through the nflfastR/nflverse dataset, which can be accessed via an R/Python workflow and downloaded in a structured table format. It is tractable because each observation is a single play with standardized fields (down, distance, yard line, time remaining, score differential, and EPA/WPA), and the dataset includes consistent game and team identifiers that support filtering and modeling. The scope is large enough to produce reliable estimates but can be subset by season and stored locally, keeping computation manageable on a personal laptop.
+I will obtain this data by installing the nflfastR package into python, and loading in the relevant data timeline. Relevant fields can then be selected from the data source and used as needed in the model.
+I will perform a logistic regression on the data, creating a binary for if the offense either goes for it or kicks it on fourth down. I can include predictors such as yards to go, current yard line, weather, time remaining, timeouts remaing, team strength, etc. This can give direct insight into what drives aggresiveness in decision making. We can also look at Expected points based on the play (calculated by the team behind the dataset) and factor that into the model as well.
+This project helps show which teams make aggressive or conservative 4th-down decisions, and whether those choices tend to help or hurt outcomes. Coaches and front offices could use it to check if their strategy matches what the numbers suggest in different game situations. Broadcasters and fans can use it to talk about decisions with evidence instead of hindsight. Overall, it gives a clearer, data-based view of a part of football that can swing games.
+Even with public data, the results can be misunderstood or used unfairly. Calling a coach “bad” based on a model can be misleading if you ignore context, small samples, or uncertainty, so the findings should be presented carefully. The work could also indirectly support sports betting, so it should be framed as analysis rather than advice. Finally, you should follow the data source’s terms of use and avoid sharing any restricted raw data.
+ + \ No newline at end of file diff --git a/Ideas/Idea_1.md b/Ideas/Idea_1.md new file mode 100644 index 0000000..eb4ef48 --- /dev/null +++ b/Ideas/Idea_1.md @@ -0,0 +1,13 @@ +Mini Project Idea 1 + +My idea is centered around the current day NFL and the analytics behind "going for it" on fourth down. The project will center around this question: "Do NFL teams go for it on fourth down in the right scenarios. What teams are most aggresive in their play calling?" + +The data is retrievable because NFL play-by-play data is available publicly through the nflfastR/nflverse dataset, which can be accessed via an R/Python workflow and downloaded in a structured table format. It is tractable because each observation is a single play with standardized fields (down, distance, yard line, time remaining, score differential, and EPA/WPA), and the dataset includes consistent game and team identifiers that support filtering and modeling. The scope is large enough to produce reliable estimates but can be subset by season and stored locally, keeping computation manageable on a personal laptop. + +I will obtain this data by installing the nflfastR package into python, and loading in the relevant data timeline. Relevant fields can then be selected from the data source and used as needed in the model. + +I will perform a logistic regression on the data, creating a binary for if the offense either goes for it or kicks it on fourth down. I can include predictors such as yards to go, current yard line, weather, time remaining, timeouts remaing, team strength, etc. This can give direct insight into what drives aggresiveness in decision making. We can also look at Expected points based on the play (calculated by the team behind the dataset) and factor that into the model as well. + +This project helps show which teams make aggressive or conservative 4th-down decisions, and whether those choices tend to help or hurt outcomes. Coaches and front offices could use it to check if their strategy matches what the numbers suggest in different game situations. Broadcasters and fans can use it to talk about decisions with evidence instead of hindsight. Overall, it gives a clearer, data-based view of a part of football that can swing games. + +Even with public data, the results can be misunderstood or used unfairly. Calling a coach “bad” based on a model can be misleading if you ignore context, small samples, or uncertainty, so the findings should be presented carefully. The work could also indirectly support sports betting, so it should be framed as analysis rather than advice. Finally, you should follow the data source’s terms of use and avoid sharing any restricted raw data. diff --git a/presentations/TestPresentation.Rmd b/presentations/TestPresentation.Rmd new file mode 100644 index 0000000..86d7236 --- /dev/null +++ b/presentations/TestPresentation.Rmd @@ -0,0 +1,16 @@ +--- +title: "Test Presentation" +subtitle: "⚔