Analysis of holding calls by play type

gistfile1.txt

###################################################
#### Analysis of holding calls by play type
#### Data from nfldb by burntsushi
#### https://github.com/BurntSushi/nfldb
#### Josh Hermsmeyer 9/20/18
###################################################

library(tidyverse)

### Load the play descriptions and play types
plays <- read_csv("data/all_plays.csv")

### Filter out all non penalty plays for now, find Offensive Holding penalties,
### then tag plays by pattern matching strings by play type: punt, scramble, pass.
### Remove holding calls on punts.
penalties <- plays %>%
   filter(note == "PENALTY") %>%
   mutate(holding = ifelse(str_detect(description, "Offensive Holding"), 1, 0)) %>%
   filter(holding == 1) %>%
   mutate(punt = ifelse(str_detect(description, "punts"), 1, 0)) %>%
   filter(punt == 0) %>%
   mutate(pass = ifelse(str_detect(description, "pass"), 1, 0)) %>%
   mutate(scramble = ifelse(str_detect(description, "scrambles"), 1, 0))

### Assume all other plays are runs. A quick check appears to confirm this.
run_penalties <- penalties %>%
   filter(pass == 0,
          scramble == 0)

### Holding calls on passing plays
pass_penalties <- penalties %>%
   filter(pass == 1)

### Holding calls on scrambles
scramble_penalties <- penalties %>%
   filter(scramble == 1)

### Add the counts together and calculate percentage of holding penalties on runs
count(run_penalties) / (count(run_penalties) + count(pass_penalties) + count(scramble_penalties))

### 59.6 %

### According to https://www.pro-football-reference.com/years/NFL/index.htm 
### league average runpass split since 2009 is ~ 45% run 55% pass