NFLData.jl
A package for intelligently loading NFL data into Julia.
NFLData.jl is a low-level data loader, designed to be a native Julia implementation of the popular {nflreadr} package. NFLData.jl makes a number of NFL data resources available quickly and handily in DataFrame format, while intelligently caching and updating these data sources to accomodate in-season changes.
Installation
NFLData.jl is available from the Julia package registry, and can be installed with the following one liner.
using Pkg; Pkg.add("NFLData")You can also add the package using the Pkg.jl REPL. Open an interactive Julia session, then press ] to open the REPL, then run add NFLData.
Getting Started
After installation, return to the Julia repl, and type:
julia> using NFLDataYou can now load data into your Julia environment. For example, to pull in NFL schedules since 1999, you can use load_schedules():
julia> load_schedules()
6978×46 DataFrame
Row │ game_id season game_type week gameday weekday gametime away_te ⋯
│ String15 Int64 String3 Int64 Date String15 Time String3 ⋯
──────┼─────────────────────────────────────────────────────────────────────────────────────
1 │ 1999_01_MIN_ATL 1999 REG 1 1999-09-12 Sunday missing MIN ⋯
2 │ 1999_01_KC_CHI 1999 REG 1 1999-09-12 Sunday missing KC
3 │ 1999_01_PIT_CLE 1999 REG 1 1999-09-12 Sunday missing PIT
4 │ 1999_01_OAK_GB 1999 REG 1 1999-09-12 Sunday missing OAK
5 │ 1999_01_BUF_IND 1999 REG 1 1999-09-12 Sunday missing BUF ⋯
6 │ 1999_01_SF_JAX 1999 REG 1 1999-09-12 Sunday missing SF
7 │ 1999_01_CAR_NO 1999 REG 1 1999-09-12 Sunday missing CAR
8 │ 1999_01_NE_NYJ 1999 REG 1 1999-09-12 Sunday missing NE
9 │ 1999_01_ARI_PHI 1999 REG 1 1999-09-12 Sunday missing ARI ⋯
10 │ 1999_01_DET_SEA 1999 REG 1 1999-09-12 Sunday missing DET
11 │ 1999_01_BAL_STL 1999 REG 1 1999-09-12 Sunday missing BAL
⋮ │ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋱
6969 │ 2024_18_CHI_GB 2024 REG 18 2025-01-05 Sunday 13:00:00 CHI
6970 │ 2024_18_JAX_IND 2024 REG 18 2025-01-05 Sunday 13:00:00 JAX ⋯
6971 │ 2024_18_SEA_LA 2024 REG 18 2025-01-05 Sunday 13:00:00 SEA
6972 │ 2024_18_LAC_LV 2024 REG 18 2025-01-05 Sunday 13:00:00 LAC
6973 │ 2024_18_BUF_NE 2024 REG 18 2025-01-05 Sunday 13:00:00 BUF
6974 │ 2024_18_MIA_NYJ 2024 REG 18 2025-01-05 Sunday 13:00:00 MIA ⋯
6975 │ 2024_18_NYG_PHI 2024 REG 18 2025-01-05 Sunday 13:00:00 NYG
6976 │ 2024_18_CIN_PIT 2024 REG 18 2025-01-05 Sunday 13:00:00 CIN
6977 │ 2024_18_NO_TB 2024 REG 18 2025-01-05 Sunday 13:00:00 NO
6978 │ 2024_18_HOU_TEN 2024 REG 18 2025-01-05 Sunday 13:00:00 HOU ⋯
39 columns and 6957 rows omittedYou can load nflfastR play-by-play data into Julia with load_pbp(years). The function can take either one or multiple years as arguments, or you can pass true into the function to pull in all years of PBP data.
julia> load_pbp(2023)
49665×372 DataFrame
Row │ play_id game_id old_game_id home_team away_team season_type week ⋯
│ Float64? String? String? String? String? String? Int32? ⋯
───────┼────────────────────────────────────────────────────────────────────────────────────
1 │ 1.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1 ⋯
2 │ 39.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
3 │ 55.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
4 │ 77.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
5 │ 102.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1 ⋯
6 │ 124.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
7 │ 147.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
8 │ 172.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
9 │ 197.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1 ⋯
10 │ 220.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
11 │ 245.0 2023_01_ARI_WAS 2023091007 WAS ARI REG 1
⋮ │ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋱
49656 │ 4684.0 2023_22_SF_KC 2024021100 KC SF POST 22
49657 │ 4709.0 2023_22_SF_KC 2024021100 KC SF POST 22 ⋯
49658 │ 4734.0 2023_22_SF_KC 2024021100 KC SF POST 22
49659 │ 4771.0 2023_22_SF_KC 2024021100 KC SF POST 22
49660 │ 4759.0 2023_22_SF_KC 2024021100 KC SF POST 22
49661 │ 4791.0 2023_22_SF_KC 2024021100 KC SF POST 22 ⋯
49662 │ 4813.0 2023_22_SF_KC 2024021100 KC SF POST 22
49663 │ 4835.0 2023_22_SF_KC 2024021100 KC SF POST 22
49664 │ 4860.0 2023_22_SF_KC 2024021100 KC SF POST 22
49665 │ 4881.0 2023_22_SF_KC 2024021100 KC SF POST 22 ⋯
365 columns and 49644 rows omitted
julia> load_pbp(2022:2023)
99099×372 DataFrame
Row │ play_id game_id old_game_id home_team away_team season_type week ⋯
│ Float64? String? String? String? String? String? Int32? ⋯
───────┼────────────────────────────────────────────────────────────────────────────────────
1 │ 1.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1 ⋯
2 │ 43.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
3 │ 68.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
4 │ 89.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
5 │ 115.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1 ⋯
6 │ 136.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
7 │ 172.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
8 │ 202.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
9 │ 230.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1 ⋯
10 │ 254.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
11 │ 275.0 2022_01_BAL_NYJ 2022091107 NYJ BAL REG 1
⋮ │ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋱
99090 │ 4684.0 2023_22_SF_KC 2024021100 KC SF POST 22
99091 │ 4709.0 2023_22_SF_KC 2024021100 KC SF POST 22 ⋯
99092 │ 4734.0 2023_22_SF_KC 2024021100 KC SF POST 22
99093 │ 4771.0 2023_22_SF_KC 2024021100 KC SF POST 22
99094 │ 4759.0 2023_22_SF_KC 2024021100 KC SF POST 22
99095 │ 4791.0 2023_22_SF_KC 2024021100 KC SF POST 22 ⋯
99096 │ 4813.0 2023_22_SF_KC 2024021100 KC SF POST 22
99097 │ 4835.0 2023_22_SF_KC 2024021100 KC SF POST 22
99098 │ 4860.0 2023_22_SF_KC 2024021100 KC SF POST 22
99099 │ 4881.0 2023_22_SF_KC 2024021100 KC SF POST 22 ⋯
365 columns and 99078 rows omitted
julia> load_pbp(true)
1186651×372 DataFrame
Row │ play_id game_id old_game_id home_team away_team season_type week ⋯
│ Float64? String? String? String? String? String? Int3 ⋯
─────────┼──────────────────────────────────────────────────────────────────────────────────
1 │ 35.0 1999_01_ARI_PHI 1999091200 PHI ARI REG ⋯
2 │ 60.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
3 │ 82.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
4 │ 103.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
5 │ 126.0 1999_01_ARI_PHI 1999091200 PHI ARI REG ⋯
6 │ 150.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
7 │ 176.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
8 │ 197.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
9 │ 218.0 1999_01_ARI_PHI 1999091200 PHI ARI REG ⋯
10 │ 240.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
11 │ 260.0 1999_01_ARI_PHI 1999091200 PHI ARI REG
⋮ │ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋱
1186642 │ 4104.0 2024_02_TB_DET 2024091503 DET TB REG
1186643 │ 4130.0 2024_02_TB_DET 2024091503 DET TB REG ⋯
1186644 │ 4155.0 2024_02_TB_DET 2024091503 DET TB REG
1186645 │ 4162.0 2024_02_TB_DET 2024091503 DET TB REG
1186646 │ 4187.0 2024_02_TB_DET 2024091503 DET TB REG
1186647 │ 4210.0 2024_02_TB_DET 2024091503 DET TB REG ⋯
1186648 │ 4233.0 2024_02_TB_DET 2024091503 DET TB REG
1186649 │ 4256.0 2024_02_TB_DET 2024091503 DET TB REG
1186650 │ 4279.0 2024_02_TB_DET 2024091503 DET TB REG
1186651 │ 4301.0 2024_02_TB_DET 2024091503 DET TB REG ⋯
366 columns and 1186630 rows omittedNFLData.jl is designed to quickly load large datasets into memory. Here, we load a clean Julia session and pull in all seasons of PBP data, containing 362 columns and, as of the time of writing this documentation, 1,186,638 plays.
julia> @time pbp = load_pbp(true);
37.171214 seconds (59.72 M allocations: 10.669 GiB, 15.75% gc time, 24.79% compilation time: <1% of which was recompilation)
Not bad for running on my pretty dinky work laptop (16 GB RM, Intel Core i7-1270P processor)!
Caching
NFLData.jl relies on Julia's JIT compilation to speed up running large objects into memory. However, JIT compiliation does not persist across sessions. NFLData.jl uses Scratch.jl to cache the data objects referenced across sessions, so subsequent calls that reference the same data objects are faster even if the call comes from a different Julia session. To learn more about this behavior, please visit the Caching chapter of this docomentation.
Logo Attribution
The NFLData.jl logo is sourced from Font Awesome Free 5.2.0 by @fontawesome. It has been remixed from its original form under the Creative Commons Attribution 4.0 International license.