Finding a way to tally soccer/football passes in R - r

Question

I am trying to find a way to take a sequence of passes and show how many times each player passes to another player.
So for example, if the pass sequence was: Jordan to Emma to Molly to Emily bad, that means Jordan's and Emma's passes were successful but Molly's was not.
I have an example of a few lines of data I put in R (in a 2x2 dataset):
Passes
1 jordan to karlie karlie turnover unforced
2 jlin to gray bad
3 alia to kiersten to lilly to kiersten bad
4 mandy to karlie bad
5 kelsey to mccarter to jordan to emma emma fouled
6 mandy to karlie bad
7 mandy to kiersten cleared
I am trying to come up with a way that can convert those lines into a table like this:
Players Mandy-G Jlin-G Gray-G Kiersten-G Kelsey-G Karlie-G Jordan-G Lilly-G Mccarter-G Emma-G Alia-G Mandy-B Jlin-B Gray-B Kiersten-B Kelsey-B Karlie-B Jordan-B Lilly-B Mccarter-B Emma-B Alia-B
Mandy 1 2
Jlin 1
Gray
Kiersten 1
Kelsey 1
Karlie
Jordan 1 1
Lilly 1
McCarter 1
Emma
Alia 1
*I don't know how to insert a screenshot, so the copy and paste messed up the formatting but you can still get the idea of what I want it to look like.
If Mandy passed to Gray and it was good there should be a 1 in the Mandy and Gray-G intersection. If Mandy passed to Gray and it was bad there should be a 1 in the Mandy and Gray-B intersection.
There are only numbers in that table because I did it by hand and it was only for about 10 minutes of a game. Ultimately, doing it for the full 90 minutes and for about 25 games, I'm going to need to create a way to go through the first table and have R sort and add a mark for each successful and unsuccessful pass.
dat3 <- strsplit(dat[,1], "to")
numPass <- rep(0, length(dat3))
for (i in 1:length(dat2)) {
temp <- sum(dat2[[i]] == "to")
if ("bad" %in% dat2[[i]]) {
temp <- temp-1
}
numPass[i] <- temp
}
maxPass <- max(numPass)+1
#for (i in 1:length(dat2)){
for (i in 5){
keep<-dat2[[i]]%in%roster[,1]
pls<-dat2[[i]][keep]
#add statemets to remove last name if there is a "bad"
for (j in 1:length(pls)) {
cols<-which(substr(names(seqPass),1,nchar(pls[j]))==pls[j])
seqPass[i,cols[j]]<-j
}
}
seqPass[c(1,5),]
I have tried the above code to go through the first five lines and to count the number of passes in each sequence and it adds a mark under each player's name if they were involved in the pass, but if it was bad they need to be removed which is does not do.
Is there a way for R to automatically count if the first name and second name in the sequence have a good pass, add a mark in their intersection, and do the same for if the first and second name make a bad pass by having the word "bad" follows the second name?
Any help would be much appreciated!
Thanks
Sample data
structure(list(VT = c("jordan to karlie karlie turnover unforced",
"jlin to gray bad", "alia to kiersten to lilly to kiersten bad",
"mandy to karlie bad", "kelsey to mccarter to jordan to emma emma fouled",
"mandy to karlie bad", "mandy to kiersten cleared bad")), row.names = c(NA,
7L), class = "data.frame", na.action = structure(8:19, .Names = c("8",
"9", "10", "11", "12", "13", "14", "15", "16", "17", "18", "19"
), class = "omit"))

Answer 1

You could use regular expressions. And Also it will be fast if you only put the data of those who touched the ball. So something like:
pass = sub('_$','_good',sub("(.*\\w+ to (?:\\w+(?=.*(bad))|\\w+)).*",'\\1_\\2',dat$VT,perl = T))
pass1 = gsub('(to(\\s[^_ ]+(?=\\s)))','\\1_good\n\\2',pass,perl=T)
results = xtabs(V3~.,cbind(read.csv(text=gsub('to',',',pass1),h=F,strip.white = T),V3=1))
results
V2
V1 emma_good gray_bad jordan_good karlie_bad karlie_good kiersten_bad kiersten_good lilly_good mccarter_good
alia 0 0 0 0 0 0 1 0 0
jlin 0 1 0 0 0 0 0 0 0
jordan 1 0 0 0 1 0 0 0 0
kelsey 0 0 0 0 0 0 0 0 1
kiersten 0 0 0 0 0 0 0 1 0
lilly 0 0 0 0 0 1 0 0 0
mandy 0 0 0 2 0 1 0 0 0
mccarter 0 0 1 0 0 0 0 0 0

Answer 2

It seems that you have done a lot of the work already, so I will just add in my two cents. It would make your table generally smaller if you didn't separate out good and bad as two tables. You could generally have one table with combinations of players like you have created, but add a column with a 1 or 0 stating if the pass was good or bad, in which case you could just have your code above but with
dat$pass <- as.numeric(grepl(".*(bad)", dat$VT))
This adds a column with 1 if the row has 'bad' in it. Imagine the complexity of a good and bad table over multiple decades and different players!