In zamorarr/msf: API for mysportsfeeds

msf is an R wrapper for the MySportsFeeds API.

There is an official R package provided by MySportsFeeds, but this package differs in the following ways:

• Provides parse_* functions to convert a JSON list into a tidy dataframe.
• Provides delay parameter in API calls to execute multiple queries with a delay between them in order to obey rate limits.
• Provides common defaults for some parameters (ex. uses season = "current") to reduce typing.

These additions have been very helpful for me to use the data effectively in R. But, if you'd rather use the official package I strongly encourage you do so. Also, when I started this package the official R package had not been released yet.

Install

devtools::install_github("zamorarr/msf")

Usage

To start, make sure you have set the following environment variables set on your computer. Putting them in your .Renviron file is probably the easiest way to do it.

MYSPORTSFEEDS_USER=YOUR_USERNAME
MYSPORTSFEEDS_PASSWORD=YOUR_PASSWORD

Then in R you can query the feeds easily. See the function documentation for a description of the parameters. They should follow the same parameters required to query the official web API.

library(msf)

# Get data
resp <- game_boxscore("nba", "20171017-BOS-CLE", season = "2017-2018-regular")

library(msf)
library(httptest)
library(withr)

# I'm not showing this part because I'm actually mocking the API call. No need to confuse the user.
#json <- read_msf("../tests/testthat/api.mysportsfeeds.com/v1.1/pull/nba/2017-2018-regular/game_boxscore.json-49e732.json")
mock_testdir <- function(expr) {
  envvars <- c("MYSPORTSFEEDS_USER" = "fakeuser", "MYSPORTSFEEDS_PASSWORD" = "fakepass")
  testsdir <- "../tests/testthat"
  with_dir(testsdir, with_envvar(envvars, with_mock_API(expr)))
}

resp <- mock_testdir(game_boxscore("nba", "20171017-BOS-CLE", season = "2017-2018-regular"))

# You can directly work with the json content represented as a list
json <- resp$content
str(json,2)

# Or parse the json list into tidy dataframe
boxscore <- parse_boxscore(json)
boxscore[1:10,1:6]

MSF Object

A response from an API call returns an msf_api object. This object is a list with four variables:

• response The entire response http response received from the MySportsFeeds API.
• path The URL used to query the data.
• content The JSON content converted into a list.
• name A default name given to the query.

Most users will likely only care about the content object. This object is a list representing the JSON data from the API call. You can view its contents with str.

json <- resp$content
str(json, 2)

The data in the JSON varies by the type of API call made. See the official docs for a description of each.

Multiple Queries

You can also run multiple queries sequentially by passing in a vector of inputs for certain fields. This will return a list of msf_api objects. You can set the delay parameter to force a delay between queries. This is to ensure that you obey the rate limits, which are currently limited to 250 requests over a 5 minute span. This comes out to approximately one query a second, which is the default delay. See the function documentation for details on which inputs can be vectors in order to run multiple queries.

resp <- game_boxscore("nba", c("20171026-BOS-MIL", "20171109-LAL-WAS"), delay = 1)
resp

resp <- mock_testdir(game_boxscore("nba", c("20171026-BOS-MIL", "20171109-LAL-WAS"), delay = 1))
resp

Queries by Date

The following functions query the MySportsFeeds API by date. If you do not provide a date, the current date will be used. If you provide multiple dates the queries will be issued sequentially with a delay between them.

• current_season
• daily_dfs
• daily_game_schedule
• daily_player_stats
• roster_players
• scoreboard

Queries by Game

The following functions query the MySportsFeeds API by game. You must provide a game id such as "40265" or "20171119-MIA-TOR". If you provide multiple dates the queries will be issued sequentially with a delay between them.

• game_boxscore
• game_pbp
• game_starting_lineup

Queries by Season

The following functions query the MySportsFeeds API by season. If you do not provide a season, the current season will be used. If you provide multiple seasons the queries will be issued sequentially with a delay between them.

• active_players
• conference_team_standings
• cumulative_player_stats
• division_team_standings
• full_game_schedule
• latest_updates
• overall_team_standings
• player_injuries
• playoff_team_standings

Queries by Team

The following functions query the MySportsFeeds API by team. You must provide a team id such as "91" or "ATL". If you provide multiple teams the queries will be issued sequentially with a delay between them.

• team_gamelogs
• player_gamelogs

Parsing JSON to Tidy Dataframe

There are parse_* functions available to convert the JSON list into a tidy dataframe. These are included for convenience and are designed to NOT include all the data in the JSON. Instead, data that is duplicated in other feeds (such as player height/weight) is silently dropped. This is to provide the simplest representation in a data frame. In some cases columns are represented as nested lists when there are many ways to parse the data. The parse functions all require you to pass the resp$content or provide a list of JSON data.

resp <- game_boxscore("nba", "20171017-BOS-CLE", season = "2017-2018-regular")
parse_boxscore(resp$content)

resp <- mock_testdir(game_boxscore("nba", "20171017-BOS-CLE", season = "2017-2018-regular"))
parse_boxscore(resp$content)

zamorarr/msf documentation built on May 3, 2019, 9:01 p.m.