In kimjam/dfstoolkit: Functions to clean fantasy sports data and make weekly projections.

library(dfstoolkit)

Starting from Scratch

Before scraping NFL box score statistics, you will need to set up a SQLite database that has two tables:

• nfl_teams
  • team_long - Full team name (e.g. Philadelphia Eagles)
  • team_abbr - 3 Letter abbreviation (e.g. PHI)
• nfl_week_date
  • week - NFL week
  • date - Dates that correspond to NFL week
  • season - NFL season

You should create the nfl_week_date table before the start of the season so that it's information is available when you are scraping the current season.

Once these two tables are set up, make sure you have DFS_DB set up in your .Renviron file.

I populated nfl_teams manually, although I'm sure you could scrape it. You can create the nfl_week_date table for 2012 - 2016 by:

library(dfstoolkit)

week_date <- lapply(2012:2016, function(x) get_nfl_week_dates(year = x)) %>%
    dplyr::bind_rows()

dfs_insert(table = 'nfl_week_date', df = week_date, overwrite = TRUE)

Now, to scrape the boxscore statistics from 2012 - 2016:

off_list <- vector('list', length(2012:2016))
def_list <- vector('list', length(2012:2016))
for (year in 2014:2016) {
  links <- get_nfl_boxscore_links(year)
  stats <- lapply(links, get_nfl_boxscore_stats)

  off_list[[match(year, 2012:2016)]] <- dplyr::bind_rows(
      lapply(stats, `[[`, 'off_stats')
  )

  def_list[[match(year, 2012:2016)]] <- dplyr::bind_rows(
      lapply(stats, `[[`, 'def_stats')
  )
}
offense <- dplyr::bind_rows(off_list)
defense <- dplyr::bind_rows(def_list)

dfs_insert(table = 'nfl_offense', df = offense, overwrite = TRUE)
dfs_insert(table = 'nfl_defense', df = defense, overwrite = TRUE)

off <- dfs_query(query = 'select * from nfl_offense limit 10;')
def <- dfs_query(query = 'select * from nfl_defense limit 10;')

The offense table will look like:

dplyr::glimpse(off)

The defense table will look like:

dplyr::glimpse(def)

Adding Data from Previous Week

During the season, you'll want to add data from the previous week into your database. For example, if you want to add week 1 data from the 2016 season, you would run this code:

links <- get_nfl_boxscore_links(year = 2016, week = 1)
stats <- lapply(links, get_nfl_boxscore_stats)

offense <- dplyr::bind_rows(lapply(stats, `[[`, 'off_stats'))
defense <- dplyr::bind_rows(lapply(stats, `[[`, 'def_stats'))

# append instead of overwrite
dfs_insert(table = 'nfl_offense', df = offense, append = TRUE)
dfs_insert(table = 'nfl_defense', df = defense, append = TRUE)

kimjam/dfstoolkit documentation built on May 20, 2019, 9:40 a.m.