War date: 2023-03-30 — latest 200 tweets
@BillGates and @elonmusk are going to war. In a fierce battle of tweets, retweets and likes, these twitter accounts have been put against each other. This dashboard showcases the results. First time at a Tweetwar? Click the WTF button below to understand what the hell is this all about.
Tweetwars is a service that allows users to compare two Twitter accounts by analyzing their latest 200 tweets and presenting the results in an easy-to-use fully interactive dashboard.
This section provides basic statistical information about the tweeting habits of both users. The statistics include the number of tweets, followers, following, retweets, likes, and average engagement rate. The dashboard allows users to quickly compare the tweeting habits of both accounts and identify any significant differences. Emotion Analysis :
Using deep learning network, this section provides an analysis of the emotional content of each user's tweets. The dashboard displays a breakdown of different emotional scores for the tweets of each account. Users can gain insight into the emotional tone of each user's tweets and identify any significant differences between them. Topic Analysis :
This section uses natural language processing techniques to identify the topics that each user tweets about. The dashboard displays the most frequently mentioned words or phrases and allows users to quickly compare the topics that each user focuses on. This can help identify similarities or differences in the interests of both users. Network Analysis :
This section provides a network analysis of how each user interacts with their network. The dashboard displays the most frequently mentioned accounts, and allows users to identify any significant differences in the way both users interact with their followers. Overall, Tweetwars provides a comprehensive analysis of two Twitter accounts, allowing users to gain insight into the tweeting habits, emotional tone, topics, and network interactions of each user. The service is useful for social media professionals, researchers, and anyone interested in understanding and comparing Twitter users.
Twitter is divided into two camps: the meticulous planners and the impulsive improvisers. The planners spend hours crafting the perfect tweet while the improvisers fire off whatever comes to mind. It's like a battle between a chess master and a freestyle rapper. Let’s check out what are @BillGates’s and @elonmusk’s tweeting habits.
Tweeting time of the day
People tweet at different times, that's for sure. Some open twitter everyday at 6pm like a clock, and others prefer to tweet along the day just as they breathe. Here you have for each contender, a bar chart with the number of tweets sent per hour of the day (hours are in UTC). The chart is interactive, you can filter by tweets / retweets / replies just by clicking on the legend.
It’s also interesting to see if twitter usage changes from weekdays to weekends, or if other patterns arise regarding the type of tweets that are written. With the following charts it’s the same story, but instead of hours we see the days of the week in which tweets are published. Again, fully interactive.
Twitter, the platform where keyboard activists come to express their deepest emotions and opinions. It's where the world comes to argue, bicker, and show off their moral superiority. A place where people use their emotions as a weapon, to gain attention, followers, or to push their agenda. But amidst all the noise, there are still genuine emotions being expressed, real people pouring their hearts out for the world to see. It's not always easy to tell the difference, but it's there, and it's worth acknowledging. It turns out that using AI we can measure how emotional each tweet is . There are five emotions measured: anger, joy, sadness, surprise and fear . There are five emotions measured: anger, joy, sadness, surprise and fear; and each tweet has a score from 0 to 1 on each emotion.
How emotional are tweets on average?
Next, you see a spider chart with the score for each emotion averaged over all tweets. You can get a quick idea of who tends to lose his temper and who is easily surprised or having the blues. Your emotions might betray you when you least expect it…
Emotions over time
Let’s now put @BillGates and @elonmusk against each other once again. In the following chart, we compare how their emotions vary through time. For the analytics nerds, it’s the weekly moving average of each emotion. Note that since we are displaying the latest 200 tweets from each user, the time windows will be different. Click on the buttons to change emotion (kind of obvious, but just in case). You can also pan through the time frame below for a more detailed view.
Most emotive tweets
Ok, it’s all fun and games but you want actual examples. Well, here you go: the most emotional tweet for each category, with its score. Notice the tiny “Tweet” button which takes you directly to the source (unless it has been deleted by an ashamed author). Remember to come back though since there is more cool stuff down the page.
This chart is for the stats geeks. It compares the percentiles of each emotion score for both twitter accounts.
For the laymen: percentiles tell us what percentage of the data is below a certain value. We first rank the tweets from lowest to highest score. Then, we plot a graph with the tweet score in the y-axis and the ranking it has (as a percentage) in the x axis.
A pair of examples in case it helps:
- Half of BillGates’s tweets (percentile 50) have a joy score below
- 90% of elonmusk's tweets (percentile 90) have a fear score below You can also toggle the scale of the y-axis between linear and logarithmic to see things more clearly. Or get even more confused.
What do people tweet about? It's no secret that Twitter users tend to have their favourite topics. Whether it's politics scandals, sports, or the latest pop culture memes, each user has their own unique interests and passions. Some use Twitter as a platform to express their opinions and share their personal experiences, while others use it as a way to showcase their expertise on a particular subject. Using a fancy clustering technique together with a deep learning language model we can extract the most common topics present in a corpus of text. Remember it’s math, not magic, so the topics found might make total sense or be absolutely nuts. Blame the person tweeting, not the AI.
These are the main topics found on the tweets of @BillGates and @elonmusk. Within each bubble there are representative terms for each topic. The bigger the bubble, the more relevant it is.
Twitter is like a massive web of connections and interactions between users. Every tweet, like, and retweet is a new node in this crazy graph that we all contribute to. Some people’s tweets are seen by millions, others do not get a single view. It all depends on “The Algorithm”. This section includes graphs that demonstrate the network effects of our users, such as the reach of their tweets, how they engage with other accounts, and more.
Who are the users which @BillGates and @elonmusk interact with the most? And what is the average anger score of the tweets mentioning those accounts? We have now the answers to these burning questions! Here comes an awesome bar chart that shows the top 5 accounts that our players mention, retweet or reply to the most. The “tweets" button displays the total tweet count, while selecting an emotion will reveal the average score for tweets mentioning that account. You might find someone who only receives tweets of happiness, or someone who gets exactly the opposite 😈.
Correlations with tweet performance
Is there any relationship between how joyful a tweet is and how many people retweet it? What about the hour in which it is tweeted and the number of likes it gets? In this correlation chart, you can check the correlation these emotions have with the tweets performance in terms of retweets, likes and replies from other users.
For each account, the x-axis represents variables such as emotions, while the y-axis represents its performance. Note that correlation indexes go from -1 to 1. High values like 0.8 or -0.9 imply stronger relationships. Negative correlations indicate inverse relationships.
Some examples follow:
- A correlation of 0.7 between “sadness” and “retweets” means that the sadder tweets are, the more retweets they get.
- A correlation of 0 between “surprise” and “replies” means that there is no direct relationship between how surprising tweets are and the number of replies they receive.
- A correlation of -0.3 between “anger” and “likes” means that angrier tweets tend to get fewer likes.