Change in average AFL game scores from 1897 to 2018

The Data

AFLTables.com provides historic data on every VFL/AFL game ever played, all the way back to the leagues inception in 1897.

The web-link can be found here: https://afltables.com/afl/stats/biglists/bg3.txt.

In this rpub, I will visualise change in average match-day scores from 1897 to the present. The approach herein will hopefuly encourage further analyses

First, lets load-in our dependencies :

library(readr)
library(tidyr)
library(dplyr)
library(anytime)
library(ggplot2)
library(plotly)
library(scales)
library(reshape)
library(data.table)
library(knitr)
library(kableExtra)

Read URL to R, specifying columns and delimiters

#read in URL
url <- 'https://afltables.com/afl/stats/biglists/bg3.txt'

#specify columns
AllAFLData <- read_table(url, col_names = c("ID", "Date", "Round", "HomeTeam",  
                                            "HomeScore", "AwayTeam", "AwayScore", 
                                            "Venue"), 
                         col_types = NULL, skip = 2)

#Convert to dataframe 
AllAFLData <- data.frame(as.list(AllAFLData))
AllAFLData <- data.frame(AllAFLData, stringsAsFactors = FALSE)

We should now have a dataframe with 8 variables, and 15,407 rows

ID	Date	Round	HomeTeam	HomeScore	AwayTeam	AwayScore	Venue
1	8-May-1897	R1	Fitzroy	6.13.49	Carlton	2.4.16	Brunswick St
2	8-May-1897	R1	Collingwood	5.11.41	St Kilda	2.4.16	Victoria Park
3	8-May-1897	R1	Geelong	3.6.24	Essendon	7.5.47	Corio Oval
4	8-May-1897	R1	South Melbourne	3.9.27	Melbourne	6.8.44	Lake Oval
5	15-May-1897	R2	South Melbourne	6.4.40	Carlton	5.6.36	Lake Oval
6	15-May-1897	R2	Essendon	4.6.30	Collingwood	8.2.50	East Melbourne

Cleaning and Feature Extraction

Change variable types

#Change variable types
AllAFLData$HomeScore <- as.character(AllAFLData$HomeScore)
AllAFLData$HomeTeam <- as.character(AllAFLData$HomeTeam)
AllAFLData$AwayTeam <- as.character(AllAFLData$AwayTeam)
AllAFLData$Date <- as.character(AllAFLData$Date)

Extract out ‘Year’ from our date column

#Duplicate date column, and separate date into day, month and year columns
AllAFLData$Date2 = AllAFLData$Date

AllAFLData <- separate(AllAFLData, Date2, 
                       into = c("Day", "Month", "Year"), 
                       sep = "[-]")

AllAFLData$Day <- as.numeric(as.character(AllAFLData$Day))
AllAFLData$Year <- as.numeric(as.character(AllAFLData$Year))

Separate ‘score’ column(s) into goals, behinds and total

#Separate score column into goals, behinds, and total 
AllAFLData <- separate(AllAFLData, HomeScore, 
                                   into = c("HomeGoals", "HomeBehinds", 
                                            "HomeTotal"), 
                                   sep = "[.]")

AllAFLData <- separate(AllAFLData, AwayScore, 
                                   into = c("AwayGoals", "AwayBehinds", 
                                            "AwayTotal"), 
                                   sep = "[.]")

Change score variables from character to numeric

#check all variable types 
sapply(AllAFLData, class)

##          ID        Date       Round    HomeTeam   HomeGoals HomeBehinds 
##   "numeric" "character"    "factor" "character" "character" "character" 
##   HomeTotal    AwayTeam   AwayGoals AwayBehinds   AwayTotal       Venue 
## "character" "character" "character" "character" "character"    "factor" 
##         Day       Month        Year 
##   "numeric" "character"   "numeric"

#change newly created numeric columns to numeric 
cols.num <- c("HomeGoals","HomeBehinds", "HomeTotal", 
              "AwayGoals", "AwayBehinds", "AwayTotal")
AllAFLData[cols.num] <- sapply(AllAFLData[cols.num],as.numeric)

#confirm change
sapply(AllAFLData, class)

##          ID        Date       Round    HomeTeam   HomeGoals HomeBehinds 
##   "numeric" "character"    "factor" "character"   "numeric"   "numeric" 
##   HomeTotal    AwayTeam   AwayGoals AwayBehinds   AwayTotal       Venue 
##   "numeric" "character"   "numeric"   "numeric"   "numeric"    "factor" 
##         Day       Month        Year 
##   "numeric" "character"   "numeric"

Aggregate teams who have merged

#change names and merge teams 
AllAFLData[AllAFLData=="South Melbourne"] <- "South Melb/Syd"
AllAFLData[AllAFLData=="Sydney"] <- "South Melb/Syd"
AllAFLData[AllAFLData=="Fitzroy"] <- "Fitzroy/Brisbane"
AllAFLData[AllAFLData=="Brisbane Bears"] <- "Fitzroy/Brisbane"
AllAFLData[AllAFLData=="Brisbane Lions"] <- "Fitzroy/Brisbane"
AllAFLData[AllAFLData=="Kangaroos"] <- "North Melbourne"
AllAFLData[AllAFLData=="Footscray"] <- "Western Bulldogs"
AllAFLData[AllAFLData=="Western Bulldog"] <- "Western Bulldogs"
AllAFLData[AllAFLData=="GW Sydney"] <- "GWS"

Finally, create an ‘era’ variable which breaks down games by the decade they occurred in. We will use this variable as a slicer for the visualisation

AllAFLData$Era <- ifelse(AllAFLData$Year>=1897 & AllAFLData$Year<=1899,"1890s",
                  ifelse(AllAFLData$Year>=1900 & AllAFLData$Year<=1909,"1900s",
                  ifelse(AllAFLData$Year>=1910 & AllAFLData$Year<=1919,"1910s",  
                  ifelse(AllAFLData$Year>=1920 & AllAFLData$Year<=1929,"1920s",
                  ifelse(AllAFLData$Year>=1930 & AllAFLData$Year<=1939,"1930s", 
                  ifelse(AllAFLData$Year>=1940 & AllAFLData$Year<=1949,"1940s",
                  ifelse(AllAFLData$Year>=1950 & AllAFLData$Year<=1959,"1950s",
                  ifelse(AllAFLData$Year>=1960 & AllAFLData$Year<=1969,"1960s",
                  ifelse(AllAFLData$Year>=1970 & AllAFLData$Year<=1979,"1970s",
                  ifelse(AllAFLData$Year>=1980 & AllAFLData$Year<=1989,"1980s",
                  ifelse(AllAFLData$Year>=1990 & AllAFLData$Year<=1999,"1990s",
                  ifelse(AllAFLData$Year>=2000 & AllAFLData$Year<=2009,"2000s",
                  ifelse(AllAFLData$Year>=2010 & AllAFLData$Year<=2019,"2010s", 0)))))))))))))

Aggregation

Let’s begin building a table summarising the total number of games and total score of each team, per decade

#Total score per team, per decade for home games
EraScore <- AllAFLData %>%
  gather(HomeAway, Team, HomeTeam) %>% 
  group_by(Era, Team) %>%
  summarise(HomeTotal = sum(HomeTotal))

#Total score per team, per decade for away games
EraScore <- AllAFLData %>%
  gather(HomeAway, Team, AwayTeam) %>% 
  group_by(Era, Team) %>%
  summarise(AwayTotal = sum(AwayTotal)) %>%
  left_join(EraScore, 
            by = c("Team" = "Team", 
                   "Era" = "Era")) 

#Total number of home & away games per team 
EraScore <- AllAFLData %>%
  gather(HomeAway, Team, HomeTeam, AwayTeam) %>% 
  group_by(Era, Team) %>%
  summarise(GamesHome = sum(HomeAway == "HomeTeam"),
            GamesAway = sum(HomeAway == "AwayTeam")) %>% 
  left_join(EraScore, 
            by = c("Team" = "Team", 
                   "Era" = "Era"))

Combine home and away scores and games to get absolute totals. Determine the average score per team, per game, per decade

#Total games per team, per decade
EraScore$TotalGames <- (EraScore$GamesHome + 
                          EraScore$GamesAway)
#Total score per team, per decade
EraScore$TotalScore <- (EraScore$HomeTotal + 
                          EraScore$AwayTotal)
#Average score per team, per decade
EraScore$AverageScore <- (EraScore$TotalScore / 
                            EraScore$TotalGames)
#round to 2 decimal places
EraScore[,'AverageScore']=round(EraScore[,'AverageScore'],2)

#Re-order variables
EraScore <- EraScore[,c(1,2,3,4,7,5,6,8,9)]

View our newly developed table

Era	Team	GamesHome	GamesAway	TotalGames	AwayTotal	HomeTotal	TotalScore	AverageScore
1890s	Carlton	22	26	48	632	578	1210	25.21
1890s	Collingwood	26	26	52	1087	1126	2213	42.56
1890s	Essendon	27	25	52	979	1529	2508	48.23
1890s	Fitzroy/Brisbane	28	23	51	903	1187	2090	40.98
1890s	Geelong	27	24	51	1083	1534	2617	51.31
1890s	Melbourne	22	29	51	960	1058	2018	39.57

Visualising the data

Create a minimalistic theme for our animated plot

#minimalistic theme for visualisation 
PlotlyTheme <-
  theme(
    panel.background = element_blank(),
    panel.grid.major = element_line(color='light grey'),
    plot.title = element_text(size=8, hjust=0.5),
    axis.title.x = element_blank(),
    axis.text = element_text(size=6),
    axis.text.x = element_text(angle = 45, hjust = 1, colour="#606060"),
    axis.text.y = element_text(hjust = 1, colour="#606060"),
    axis.title = element_text(size=8),
    axis.line.x = element_line(color="black", size = 0.5),
    axis.line.y = element_line(color="black", size = 0.5),
    legend.key = element_rect(fill = "white") +
      scale_y_continuous(label = comma)
  )

Visualise the average score per match, per team, across each decade

#animated plot, summarising every teams average score per decade
animatedscore <- EraScore %>% 
  ggplot(aes(x = Team, y = AverageScore)) + 
  geom_point(aes(frame = Era), colour="#3599B8") + 
  stat_summary(aes(y = AverageScore, group = 1, frame = Era), 
               fun.y=mean, colour="#3599B8", geom = "line", group = 1) +
  labs(title = "Average score per game, per decade for all AFL teams", 
       y = "Average score") +
  PlotlyTheme

This interactive visualisation clearly shows that with each passing decade up until the 1980s, incremental increases in average score per team, per game can be observed. Scoring has increased nearly three-fold since the 1890s. From 1990 to the present, there has been a slight downward trend in overall scoring. This is probably attributable to more sophisticated defensive tactics as the game has matured into a professional competition.

Change in average AFL game scores from 1897 to 2018

perkot

07 December 2018

The Data

Cleaning and Feature Extraction

Aggregation

Visualising the data