A lag aligns each observation with an earlier observation at a chosen offset. In dplyr, lag(points) supplies the previous free-throw result so each attempt can be compared with what came immediately before it. section 16.4.1 section 16.4.2
Use lag to prepare sequential sports data:
- Group attempts by date so a previous game cannot influence the current game's streak. section 16.4.1
- Create a streak with streak_run(lag(points)); the lag makes the streak depend on the preceding record. section 16.4.1
- Filter with lag(points) == 1 to retain attempts following a made free throw, then summarize makes and misses by streak. section 16.4.1 section 16.4.2
- Remember that lead() performs the opposite operation, while lag can look back by two or more records. section 16.4.1
Comparing free throws after a make
For Giannis Antetokounmpo, group the data by date and calculate streak = streak_run(lag(points)). After removing each game's final record, filter for lag(points) == 1, count makes and misses for each streak, and calculate the resulting free-throw percentage. The same workflow produces Julius Randle's comparison table and chart. section 16.4.1 section 16.4.2
giannis %>% group_by(date) %>% mutate(streak = streak_run(lag(points))) -> giannis_final head(giannis_final, n = 10) ## # A tibble: 10 × 7 ## # Groups: date [1] ## date event_type player points result ## <date> <fct> <chr> <dbl> <fct> ## 1 2019-10-24 free throw Giannis Antetokounmpo 0 missed ## 2 2019-10-24 free throw Giannis Antetokounmpo 1 made ## 3 2019-10-24 free throw Giannis Antetokounmpo 0 missed ## 4 2019-10-24 free throw Giannis Antetokounmpo 0 missed ## 5 2019-10-24 free throw Giannis Antetokounmpo 0 missed ## 6 2019-10-24 free throw Giannis Antetokounmpo 1 made ## 7 2019-10-24 free throw Giannis Antetokounmpo 1 made ## 8 2019-10-24 free throw Giannis Antetokounmpo 1 made ## 9 2019-10-24 free throw Giannis Antetokounmpo 1 made ## 10 2019-10-24 free throw Giannis Antetokounmpo 1 made ## type streak ## <fct> <int> ## 1 Free Throw 1 of 2 0 ## 2 Free Throw 2 of 2 1 ## 3 Free Throw 1 of 2 1 ## 4 Free Throw 2 of 2 1 ## 5 Free Throw 1 of 2 2 ## 6 Free Throw 2 of 2 3 ## 7 Free Throw 1 of 1 1 ## 8 Free Throw 1 of 2 2 ## 9 Free Throw 2 of 2 3 ## 10 Free Throw 1 of 2 4
Inspect the executable dplyr pipeline that groups free throws by date and derives a streak from lagged points. section 16.4.1
