nflfastR
R package nflfastR by nflverse — release notes from GitHub.
nflfastR is shedding surface to the rest of nflverse and consolidating on one stats API.
◆Recent moves
- 6mo ago
Loaders re-exported from nflreadr; legacy scrapers deprecated
calculate_stats() gains a pbp argument so stats can be computed over a subset of play-by-play, the nflreadr loaders become direct re-exports, and fast_scraper_roster(), fast_scraper_schedules() and report() are deprecated. The consolidation started in 5.0.0 continues, with xgboost v3 compatibility and a run of data-accuracy fixes alongside.
View source ↗ - 1y ago
Standings handed to nflseedR; R 4.1 now required
calculate_standings() is deprecated in favour of nflseedR's implementation and the package now requires R 4.1 for the native pipe. Another responsibility handed to a sibling package rather than maintained here.
View source ↗ - 1y ago
nflfastR 5.0.0
⚡ SPARKcalculate_stats() unifies the whole calculate_player_stats*() family behind one faster function, with nfl_stats_variables documenting every output column and five dependencies dropped. This is the hinge the two later releases build on.
View source ↗ - 2y ago
Season-level conversion rate aggregation fixed
Corrects season-level aggregation in calculate_series_conversion_rates() and adjusts test behaviour at CRAN's request. A maintenance release.
View source ↗ - 2y ago
Raw play-by-play can now be cached and parsed locally
Adds full support for loading raw play-by-play from the local filesystem, with save_raw_pbp() and missing_raw_pbp() to populate and audit that directory. A real workflow change for anyone rebuilding seasons repeatedly.
View source ↗ - 3y ago
Reverse-dependency tests and dplyr compatibility fixes
Reverse-dependency tests, internal performance work and fixes for upcoming dplyr and tidyselect changes. Insulation against the tidyverse moving underneath it.
View source ↗