README.md

In ecastillomon/nflgisR: What the Package Does (One Line, Title Case)

nflgisR

nflgisR is a project to bring the tools needed for analyzing NFL tracking data. While this data is still not fully public, this package aims to help advancing research on NFL tracking data, and sports analytics in general. The project aims to provide a simple framework that can be followed to compute any statistic needed to characterize a play.

Tracking data is converted to simple features to facilitate computation of geometric objects. Harnessing two of R’s biggest strenghts: the tidyverse and sf, nflgisR provides a framework to handle and organize tracking data in the NFL. This project is totally new, and is open to collaboration, or extensions to different sports. It is designed to use the columns provided for the 2020 NFL Data Bowl. If you have a different dataset, you would have to set the names to match those files or some minor changes to the code in load_db. More information regarding sources is provided in the documentation for load_db

Installation

To use this package you can sign into Kaggle and download the data at www.kaggle.com/c/nfl-big-data-bowl-2021/data

This package depends on sf, which could require you to install some extra libraries for geographic data. There are some tutorials to do these on https://thinkr.fr/Installation_spatial_EN.html

Place all the data at data/

You can build a database to handle the files if you have DBI and RSQLite, or use csv’s.

devtools::install_github("ecastillomon/nflgisR")

Getting started

Run load_db to load all necessary connections. If the DB=sql, it will lazy load tables using dbplyr. csv mode is much faster but it will be much more intense on the RAM.

The main function is get_play_sf, which will load a vector of gameIds and playIds. If no playId is given, it will go for whole games. If you want to make large extractions, you should pass a vector of gameIds and playId=NULL.

library(nflgisR)
load_db(src_dir="data",DB="csv",week=5)


sf=get_play_sf(51,2018100400)
sf=get_vornoi_transformations(sf)

Once you have plays as nested dataframes, with all the information you need. Computation between columns becomes very easy. In these example convex hulls and delaunay triangles are computed for offense and defense. Checkout how this functions work, so that you can easily add any transformation you need.

sf=get_play_sf(51,2018100400) %>% 
  get_closest_players() %>% 
  get_convex_hull_off() %>% 
  get_convex_hull_def() %>% 
  get_delaunay_triangles_def() %>% 
  get_delaunay_triangles_off()

If you want to compute play-wide statistics, you can do it in a similar way, as in the example:

sf=get_play_sf(51,2018100400) %>% 
  get_play_stats()
sf$play_stats[[1]]

To create the database, run create_db, you will need to pass a src_dir and a target_dir. Database will be created using SQLite, which is very slow and heavy, so if you find yourself with some time, contact me to see if we can plug in a faster one.

create_db(src_dir = "/home/esteban/nflgisR/data/")

gganimate is not installed to make the package lighter, but you will have to install if you don’t already have it already to create gifs. Plots were used in the submission for my databowl project that can be found here https://www.kaggle.com/est092/vornoi-areas

plot_vornoi_animation(df)

plot_vornoi_analytics(df)

Thanks for using! This project is meant to be open-source, and any feedback or contribution will be considered. Feel free to contact me at ecastillomon, or at patpAItriot for more football analytics project. Email: [e-mail]

Github: https://github.com/ecastillomon

ecastillomon/nflgisR documentation built on Jan. 25, 2021, 12:35 a.m.