assembled with blockbuilder.org and htmwidgets
I had an idea (sure I'm not the only one) to visualize football or sports in general as sequences using d3.js, so I tweeted and got a whopping two likes. Not a lot of motivation, but away I went using last weekend's Alabama vs. Texas A&M game.
With some very slight data work in R on the drive-chart HTML table and the sunburstR htmlwidget, I was able to easily create a prototype. There is a whole lot of potential improvement here starting with colors and explanation. Feel free to contribute. Thanks to Kerry Rodden one more time for his sequences sunburst.
# sunburst of drive chart
# from rolltide.com
library(htmltools)
library(xml2)
library(rvest)
library(pipeR)
library(dplyr)
library(lubridate)
library(sunburstR)
library(d3r)
# Alabama vs Texas A&M
url <- "http://rolltide.com/boxscore.aspx?id=4471&path=football"
# make simple converter for minutes:seconds to seconds
# based off http://stackoverflow.com/questions/29067375/r-convert-hoursminutesseconds
minsec_to_sec <- function(tim){
res <- lubridate::ms(tim) # format to 'hours:minutes:seconds'
lubridate::minute(res)*60 + lubridate::second(res)
}
url %>>%
read_html() %>%
html_node("section#all-drives table") %>>%
html_table(fill=TRUE, header=FALSE) %>>%
{
cnm <- .[2,]
tbl <- .[-(1:2),]
colnames(tbl) <- cnm
tbl
} %>>%
{.[,c(2,3,9,11)]} %>>%
mutate(
Qtr = gsub(x=Qtr.,pattern="[a-z,A-Z]",replacement=""),
result = `How Lost`,
size = minsec_to_sec(TOP)
) %>>%
select(Team, Qtr, result, size) %>>%
d3r::d3_nest(value_cols="size") %>>%
{
sunburst(
jsondata = .,
count=TRUE,
percent=TRUE,
sortFunction=htmlwidgets::JS(
"
function(a,b){debugger;return d3.ascending(a,b)}
"
)
)
}
I did not originally plan to do a treemap, but R makes it so easy that I could not resist. See treemap of football.
to be done... See Brendan Gregg's post on flamegraphs. Please help if you are interested.
