Site icon R-bloggers

Analysis of the Time Aspect of the Matches at The International 3 (Dota 2 Tournament)

[This article was first published on Misanthrope's Thoughts, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
It’s been a while since I analysed something with R. I decided to pick into Dota 2 statistics. This game is the real eSport now with players fighting for respectable prise pools. And the main Dota 2 event is The International where best teams compete each other in astonishing battles.

So I grabbed some stats on The International 3 from here and started to think what can I do with it… The one major aspect of the game that was actively discussed prior the Tournament was the Dire side has an advantage over the Radiant via having Roshan Pit in their possession. Many people still think that Dire side is preferable in competitive games (they say that Roshan is killed by the Dire in 70% of cases). In fact The Alliance (TI3 winner) played like 90% of their games on the Dire side at the tournament. But is it a valid proof of the Dire side advantage? – I doubt. I think that in contrary – the Radiant side has an advantage over Dire, but I will add my arguments after I prove I’m right.

Ok, here is my hypothesis. There is no time limitation for the game. The match in Dota 2 lasts until the main building of the one of the sides is destroyed (or one of the team will give up). So if one of the sides (all the other things being equal) has an advantage it will cause the median time that was used to win the game for this side to be lower than to win for the other (and vice versa for time to loose). So lets take a look at the boxplot below:

Dire Vs Radiant winning time comparisson

Code:
library(ggplot2)

TI3 <- read.csv("~/R/Dota2_Stats/The_International-3")

# create winning time per side plot
p <- ggplot(TI3, aes(TI3[,5], TI3[,8])) +
  geom_boxplot(aes(fill = TI3[,5]))+
  labs(title = 'Winning time per side at TI-3',
       x = 'Side',
       y = 'Time, min.',
       fill = 'Side')
  
print(p)
Clearly, the Radiant side generally wins slightly quickly then Dire (and have higher number of wins: 82 against 76). This means that not the Dire, but the Radiant team has an advantage in game. But why? (You may skip the rest of the paragraph if you never played Dota). There are several reasons. Radiant advantages are: easier rune control, ability to deny creeps at the hard lane, camera angle (it is not parallel to the terrain surface and facing north towards the Dire side). Camera angle was never discussed as the advantage/disadvantage because most people just got used to it, but Radiant has slight, but sure vision advantage. Seems Roshan accessibility and a safer ancient camp does not help that much to the Dire. 

What else can we do with time analysis? We can compare win and loss times for all the teams that competed at TI3:
Teams winning and loosing time comparison
Code:
# get the list of teams
Teams <- unique(TI3[,3])

# create dataset to store data about winning/loosing times
A_time <- data.frame('Team'= character(),
                     'Result' = character(),
                     'Time' = numeric(),
                     stringsAsFactors = FALSE)

# extract time data and write it to the A_time data frame
for (team in Teams) {
  A <- subset(TI3, TI3[,3] == team | TI3[,4] == team)
    
  for (i in 1:nrow(A)) {
    winner <- A[i,][5]
    dire <- A[i,][4]
    radiant <- A[i,][3]
    time <- A[i,][8]
    if ( (winner == 'DIRE' & dire == team) | (winner == 'RADIANT' & radiant == team) ) {
     result <- paste(team, 'WIN')
    }
    else {
     result <- paste(team, 'LOSS') 
    }
    A_time[(nrow(A_time)+1),] <- c(team,result, time)
  }
}

# create plot for winning time per team
p <- ggplot(A_time, aes(A_time[,2], A_time[,3])) +
     geom_boxplot(aes(fill = A_time[,1]))+
     theme(axis.text.x = element_text(angle=90, hjust = 0),
          axis.title = element_text(face = 'bold', size = 14),
          title = element_text(face = 'bold', size = 16),
          legend.position = 'right'
          ) +
     labs(title = 'Win and loss times at The International 3',
          x = 'Results',
          y = 'Time, min.',
          fill = 'Teams')
  
print(p)

Generally my assumption was correct: better teams wins quicker. Alliance and NaVi  (1-st and 2-nd places) is a nice conformation to it (DK and IG (TI-2 champion) have the similar pattern as well despite shared 5-th place). But Orange and TongFu (3-rd and 4-th place) tends to loose quicker than win. This could be explained by the general playstile of this two Asian teams which often aims at the late game. This infamous style with prolonged no action farming stage is often referred as 'Chinese Dota'. But DK and IG are Chinese teams too.   Seems that both TongFu and Orange were able overcame the odds and jumped over their heads in the given tournament. They took places that DK and IG should have get (DK and IG were more favourable teams than Orange and TongFu before the tournament).

To leave a comment for the author, please follow the link and comment on their blog: Misanthrope's Thoughts.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.