The FootballData package and dashboard

Table of Contents

Intro

A colleague made the following site in jest (https://coppingr.com/), as an outgoing present for the founding head of our department. Inspired this effort, I wondered whether you could make a service to predict a football supporters mood, by leveraging one of the many APIs out there that serve live or near live football data.

This post is an example of using the package I wrote that spawned from that idea.

Documentation: https://epijim.github.io/FootballData/

Setup environment

library(FootballData)
library(tidyverse)
library(glue)
library(emo) #devtools::install_github("hadley/emo")


# table helper - switch between html vs markdown
table_it <- function(x, type = "markdown"){
 if(type == "markdown") return(x %>% knitr::kable(escape = FALSE))
 x %>% kableExtra::kbl(escape = F) %>%
   kableExtra::kable_classic("striped", full_width = TRUE)
}

Get competitions

First I will pull all the leagues that are in the free tier, from this particular API (football-data.org) and show them on a table.

# Get competitions
competitions <- football_get_competition() %>%
  filter(
    # filter to what's in the free plan
    plan == "TIER_ONE"
    &
    # The data structures returned for country level competitions is
    # different, so let's remove it
    !country %in% c("World","Europe"))

competitions %>% select(flag,country,league) %>% table_it()

flag	country	league
NA	Brazil	Série A
	England	Championship
	England	Premier League
	France	Ligue 1
	Germany	Bundesliga
	Italy	Serie A
	Netherlands	Eredivisie
	Portugal	Primeira Liga
	Spain	Primera Division

Dive into a league

Now I’ll pick my local premier league (Bundesliga) and pull info on it’s current standings.

# Get league info
league <- competitions %>%
  filter(country == "Germany" & league == "Bundesliga") %>%
  pull(competition_id) %>%
  football_get_standing()

league %>%
  mutate(team = paste(crest,team)) %>%
  select(team,playedGames,form,points) %>%
  table_it()

team	playedGames	form	points
FC Bayern München	20	W,W,W,W,W	48
RB Leipzig	20	W,W,L,W,D	41
VfL Wolfsburg	20	W,W,W,W,D	38
Eintracht Frankfurt	20	W,W,W,D,W	36
Bayer 04 Leverkusen	20	W,L,L,W,L	35
Borussia Dortmund	20	L,W,L,L,D	32
Borussia Mönchengladbach	20	L,D,W,W,D	32
SC Freiburg	20	W,L,W,D,L	30
1. FC Union Berlin	20	L,D,L,L,W	29
VfB Stuttgart	20	L,W,L,L,D	25
SV Werder Bremen	19	D,W,L,W,D	22
TSG 1899 Hoffenheim	20	L,L,W,W,D	22
FC Augsburg	20	L,L,W,L,L	22
1. FC Köln	20	W,W,L,W,D	21
Hertha BSC	20	L,L,L,L,D	17
DSC Arminia Bielefeld	19	L,L,W,D,W	17
1. FSV Mainz 05	20	W,L,W,L,D	13
FC Schalke 04	20	L,D,L,L,L	8

Get upcoming games for a team

From the API, I can also get more details on my local team.

team_of_interest <- "SC Freiburg"

league %>%
  filter(team == team_of_interest) %>%
  pull(team_id) %>%
  football_get_upcoming() %>%
  arrange(utcDate) %>%
  slice(1) %>%
  mutate(
    text = glue(
      "Today is {Sys.Date()}\n",
      "{team_of_interest}'s next game is on {as.Date(utcDate)}\n",
      "Home: {homeTeam$name}; Away {awayTeam$name}"
    )
  ) %>%
  pull(text)
#> Today is 2021-02-09
#> SC Freiburg's next game is on 2021-02-13
#> Home: SV Werder Bremen; Away SC Freiburg

Calculate the moods

Now that I have used the data prep functions, I can jump into the ‘fun’ bit of this package. Using the live data I can run simple rules to try and figure out what the mood of a supporter may be right now.

Below - I run the mood logic on the Bundesliga.

emoji_bad <- emo::ji("poop")
emoji_good <- emo::ji("beer")

add_bad <- function(x){
  gsub("-",emoji_bad,x = x)
}

add_good <- function(x){
  gsub("\\+",emoji_good,x = x)
}

league %>%
  football_calculate_competition_metrics() %>%
  mutate(
    team = paste(crest,team),
    metric_mood = as.character(metric_mood),
    metric_mood = case_when(
        str_detect(metric_mood,"Jubilant") ~ kableExtra::cell_spec(
            metric_mood, color = "black", background = "#98FB98"
        ),
        str_detect(metric_mood,"Optimistic") ~ kableExtra::cell_spec(
            metric_mood, color = "black", background = "#e5f5f9"
        ),
        str_detect(metric_mood,"Ambivalent") ~ kableExtra::cell_spec(
            metric_mood, color = "black", background = "#f7fcb9"
        ),
        str_detect(metric_mood,"Doldrums") ~ kableExtra::cell_spec(
            metric_mood, color = "black", background = "#ffc4c4"
        ),
        TRUE ~ metric_mood
    )
  ) %>%
  rename(mood = metric_mood,numeric_mood = metric_mood_numeric) %>%
  mutate_at(vars(matches("metric_")),add_bad) %>%
  mutate_all(add_good) %>%
  select(team,points,starts_with("metric_"),numeric_mood,mood) %>%
  table_it()

team	points	metric_season_position	metric_recent_wins	metric_back_of_net	numeric_mood	mood
FC Bayern München	48	🍺🍺🍺🍺 top of the league	🍺🍺🍺🍺 5 win streak	🍺🍺🍺 strikers are on form	11	Jubilant: FC Bayern München are on a roll
RB Leipzig	41	🍺🍺 strong league position	🍺🍺 recent wins	🍺🍺🍺 strikers are on form	7	Jubilant: RB Leipzig are on a roll
VfL Wolfsburg	38	🍺🍺 strong league position	🍺🍺 recent wins	🍺 scored more than conceded	5	Optimistic: VfL Wolfsburg are doing well
Eintracht Frankfurt	36	🍺 safe in middle of league	🍺🍺 recent wins	🍺 scored more than conceded	4	Optimistic: Eintracht Frankfurt are doing well
Bayer 04 Leverkusen	35	🍺 safe in middle of league	💩💩 more recent losses than wins	🍺 scored more than conceded	0	Ambivalent: Bayer 04 Leverkusen aren’t doing well, but hey - at least they aren’t FC Schalke 04
Borussia Dortmund	32	🍺 safe in middle of league	💩💩 more recent losses than wins	🍺 scored more than conceded	0	Ambivalent: Borussia Dortmund aren’t doing well, but hey - at least they aren’t FC Schalke 04
Borussia Mönchengladbach	32	🍺 safe in middle of league	drew last game	🍺 scored more than conceded	2	Optimistic: Borussia Mönchengladbach are doing well
SC Freiburg	30	🍺 safe in middle of league	💩 lost last game	🍺 scored more than conceded	1	Optimistic: SC Freiburg are doing well
1. FC Union Berlin	29	💩 so so season	🍺 won last game	🍺 scored more than conceded	1	Optimistic: 1. FC Union Berlin are doing well
VfB Stuttgart	25	💩 so so season	💩💩 more recent losses than wins	🍺 scored more than conceded	-2	Ambivalent: VfB Stuttgart aren’t doing well, but hey - at least they aren’t FC Schalke 04
SV Werder Bremen	22	💩 so so season	drew last game	💩 conceded more than scored	-2	Ambivalent: SV Werder Bremen aren’t doing well, but hey - at least they aren’t FC Schalke 04
TSG 1899 Hoffenheim	22	💩 so so season	drew last game	💩 conceded more than scored	-2	Ambivalent: TSG 1899 Hoffenheim aren’t doing well, but hey - at least they aren’t FC Schalke 04
FC Augsburg	22	💩 so so season	💩💩 more recent losses than wins	💩 conceded more than scored	-4	Ambivalent: FC Augsburg aren’t doing well, but hey - at least they aren’t FC Schalke 04
1. FC Köln	21	💩 so so season	🍺🍺 recent wins	💩 conceded more than scored	0	Ambivalent: 1. FC Köln aren’t doing well, but hey - at least they aren’t FC Schalke 04
Hertha BSC	17	💩💩 poor season	💩💩 no recent wins	💩 conceded more than scored	-5	Doldrums: Hertha BSC are doing poor. Now’s a good time to bad mouth the manager.
DSC Arminia Bielefeld	17	💩💩 poor season	🍺 won last game	💩💩💩 defense is a seive	-4	Ambivalent: DSC Arminia Bielefeld aren’t doing well, but hey - at least they aren’t FC Schalke 04
1. FSV Mainz 05	13	💩💩💩💩 relagation?	drew last game	💩💩💩 defense is a seive	-7	Doldrums: 1. FSV Mainz 05 are doing poor. Now’s a good time to bad mouth the manager.
FC Schalke 04	8	💩💩💩💩 relagation?	💩💩 no recent wins	💩💩💩 defense is a seive	-9	Doldrums: FC Schalke 04 are doing poorly. Avoid supporters at all costs.

Shiny app

Not interested in the Bundesliga? I also wrapped the package in a shiny app available here: https://epijim.shinyapps.io/football_moods/

Exactly what are the metrics that define mood?

The current logic is contained in the following Github file. As a summary, following gist highlights just the relevant code present when this post was authored.