Twitter Data Analysis for theScore eSports

Being a fan and user of the theScore, I was interested in doing some analytics for any data that I could grab. Using the TM package in R, I was able to data mine 2501 tweets from theScore eSports Twitter account. The latest tweets I obtained were on August 31st, 2016. With the data, I was curious to determine what tweets obtained a high level of user engagement.

Of the 2501 tweets, only 101 had retweets of 20 or more. One small difference I noticed was if the tweet contained a URL of a picture or video. Of the 101 top tweets, 99 had URLs leading to a 98% picture to tweet rate. For the whole dataset, 2362 contained a URL leading to a 94% picture to tweet rate. This could indicate that links may lead to more user engagement, though this is still not completely clear as the percentages are similar.

Word Frequency and and Average Retweet Rate

From the 2501 tweets, the most frequent terms used were icymi (in case you missed it) and eSports such as LoL, Dota 2 and CS Go. Even though these were the most used terms by theScore Esports in its Twitter account, did that still translate to the most retweets or user engagement?

An interesting insight follows under Overwatch. Even though Overwatch is not used as frequently as League or Dota 2, it still receives a higher average retweet count. Following Overwatch in highest average retweets is LoL and Fnatic. This might indicate that more tweets should have emphasis on Overwatch, League and Fnatic if theScore wants more user engagement. It is possible that the higher retweets for Overwatch maybe in part to its novelty, so further testing to see how the retweet count changes over time should be considered.

Frequent Terms and Retweet Average for Top Tweets

I created similar graphs for the top 101 tweets by theScore eSports to see if there were other interesting insights.

It seems that for the 101 top tweets, there is a strong emphasis on Dota 2 and its International Tournament. However even though Dota 2 is still used more than League and CS Go, it still falls behind in the amount of average retweets. Also compared to the frequent terms of the original dataset, we see a few more eSports teams and tournaments such as SK Gaming, Cologne and EVO for Super Smash Brothers. This might lead to another important take away for more user engagement. From the data, less emphasis on Dota 2 and more on big tournaments and professional teams could lead to more retweets. While this is not guaranteed, the data warrants further testing in these areas.

I hope my analysis sparks some fun discussions on further strategies to improve user engagement at theScore eSports. This was a really fun project and thinking about how analytics can improve gaming is something I am passionate about and will continue to pursue.

All the Best my fellow Gamers!

Istiak Gani