Skip to content

Instantly share code, notes, and snippets.

@friscojosh
Last active January 27, 2021 16:13
Show Gist options
  • Save friscojosh/f59e3c5b71da541a9a34ea27a53bafef to your computer and use it in GitHub Desktop.
Save friscojosh/f59e3c5b71da541a9a34ea27a53bafef to your computer and use it in GitHub Desktop.
WOPR year over year stability analysis
############################################################################
#### Grab airyards data from airyards.com and test the year-to-year
#### stability of WOPR (Weighted Opportunity Rating)
############################################################################
library(tidyverse)
library(jsonlite)
## define a function that uses jsonlite to call the airyards.com
## API and returns the processed JSON as a dataframe.
get_air_yards <- function(year){
uri <- paste0("http://api.airyards.com/", year, "/weeks")
df_air <- fromJSON(uri) %>%
mutate(season = year)
return(df_air)
}
## initialize an empty data frame so we have something to work with
wopr_data <- data.frame()
## for loops are generally frowned upon in R, but this is fine and fuck the haters.
## we take each year of data and bind all the rows to create one large dataframe.
for (year in 2009:2017) {
print(paste("Grabbing air yards data for", year))
df_air <- get_air_yards(year)
wopr_data <- rbind(wopr_data, df_air)
}
## group all the weekly data into player-seasons,
## re-calculate WOPR for each season, and add a column called season2
## to join on in the next step
wopr_data_seasons <- wopr_data %>%
filter(tar >= 1) %>%
group_by(player_id, full_name, season) %>%
summarize(team_att = sum(tm_att),
team_air = sum(team_air),
targets = sum(tar),
air_yards = sum(air_yards)) %>%
mutate(wopr = round(0.7 * (air_yards / team_air) + 1.4 * (targets / team_att), 2),
season2 = season + 1) %>%
select(player_id, full_name, wopr, season, season2) %>%
arrange(-wopr)
## join the dataframe on itself using the season + 1 column we just made
wopr_joined <- wopr_data_seasons %>%
left_join(wopr_data_seasons, by = c("player_id", 'season2' = 'season')) %>%
na.omit()
## check the correlation by creating a simple linear model
model <- lm(data = wopr_joined, wopr.y ~ wopr.x)
summary(model)
## draw a quick scatter plot to see the relationship year over year
plot(wopr_joined$wopr.x, wopr_joined$wopr.y)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment