Watson Analytics Breaks Down ‘Minions’ Social Media Buzz [Updated]

by   |   July 31, 2015 5:19 am   |   0 Comments

Suman Mukherjee, IBM Watson Analytics

Suman Mukherjee, IBM Watson Analytics

I am a movie buff. I’m also smitten by the promise of self-service analytics to empower everyone in an organization with access to data (even spreadsheets) to ask analytical questions without the need to depend either on IT or any pre-requisite expertise. Given the buzz surrounding today’s U.S. release of Minions, I wanted to know more about what people across the world have been saying about the film.

I fired up IBM Watson Analytics in the Cloud to analyze the global Twitter chatter from the past five months leading up to today’s release.

Watson Analytics is a SaaS-based product that empowers any business user with data – which could be as simple as spreadsheets/.csv files or other cloud or on-prem sources – to just start asking questions without having to depend on IT or any prerequisite expertise.

Using the Twitter Data Connector, I import the last five months’ Twitter chatter from #Minions, as shown in the following screen capture.

 

Click to enlarge

Click to enlarge

 

Once the dataset is imported, I quickly create a data group based on the author following, namely Low (<= 150) and High (> 150). This will allow me to focus only on Tweets by those who, in my opinion, have a good Twitter following. I do my data shaping in the Refine feature and can reuse these changes in all the other features (like Explore, Predict, and Assemble).

Click to enlarge

Click to enlarge

 

Now that my data is imported, I have access to information including Twitter sentiment (because the data is passed through the Social Media Analytics engine during the import), region-based fields (like author city, state, and country), time hierarchies (created automatically based on the Tweet posting time), language, the Tweet source (where the Tweet was posted from), and author-related details.

Using the refined data as my new dataset, I type a question in natural language: “I want to know about the Tweets across countries.” But even without typing a question, I am provided with relevant starting points.

Click to enlarge

Click to enlarge

 

I click on the first insight, filter on positive sentiments, and realize that the British people (the film was released in the UK on June 26th) have been raving about it.

Click to enlarge

Click to enlarge

 

I am now curious about those people who were not Tweeting favorably. I want to focus only on those who have a high following (followers > 150, based on the data group I created earlier). By the looks of it, a few Americans are not very happy about the movie.

Click to enlarge

Click to enlarge

 

I want to know more about the negative sentiments. First I’ll take a quick look at the Tweet volumes over these months.

Click to enlarge

Click to enlarge

 

Minions was released in 44 overseas markets in the month of June and July ahead of today’s North American release. This trend chart of the Tweet volume during June tells that story nicely.

Click to enlarge

Click to enlarge

 

I take a look at how the sentiments varied over the last five months.

Click to enlarge

Click to enlarge

 

Looking at only the sentiments expressed from the UK, we can see that negative sentiments plunge after the release late in June. The sentiment analysis not only is easy to understand and use, but also has huge applications in businesses of any size that are interested in tracking and understanding customer sentiment, and is relevant across industries and geographies.

Click to enlarge

Click to enlarge

 

Earlier we learned that United States had the highest negative sentiments expressed, and that’s interesting enough for me to look into. I break down the sentiments expressed across Languages only for U.S.-based Tweet authors who have a high following. I realize that the bulk of the positive sentiments were expressed in German (followed by English and Spanish). This is surely because Minions was released in Germany and Spain on July 2nd and 3rd, respectively, while Americans have had to wait for today’s release.

Click to enlarge

Click to enlarge

 

Also interesting to note is that the bulk of the positive sentiments from females in the U.S. are being expressed in Spanish.

Click to enlarge

Click to enlarge

 

Here I look into the U.S. cities across author gender to locate exactly where the bulk of the negative Tweets are coming from. I could share all the insights that I gleaned up until now as an email attachment in a format of my choice (.jpg/.ppt/.pdf) right from here if I wanted to seek my colleagues’ opinions.

Click to enlarge

Click to enlarge

 

With the above understanding, I quickly create a dashboard by dragging and dropping the fields (using the Assemble feature, where I can also use multiple datasets to create multi-tabbed dashboards). I focus my analysis on the United States and only on Tweet authors with a large following. I want to understand the source of such Tweets (“Posted From” bar chart) and also hone in on specific authors.

Click to enlarge

Click to enlarge

 

Every visualization, filter, etc., that we see is interactive. I further filter on negative sentiments and, based on my selection of the United States (since this entire dashboard is already filtered on the United States), I can easily focus on the Tweet sources (Posted From) and especially the Author Names (on the far right).

Click to enlarge

Click to enlarge

 

So, among other things, I learned that Twitter authors in the UK are raving about Minions, and that the greatest amount of negative sentiment being expressed is coming from the United States. However, I also learned that negative sentiment in the UK fell considerably after the release of the movie there, so it’s reasonable to expect that, with today’s U.S. release, this effect will be repeated in the United States.

Imagine the implications for your business when you can focus attention on the region, source, and even on specific Twitter authors. This granular level of data insight allows you to run campaigns with unprecedented detail.

And the fact that anyone in the organization can access these data insights on their own, without analytics expertise or waiting on IT, provides businesses with the agility to respond to market changes and social sentiment on the fly, in near real time. This kind of insight is available about customer groups of every stripe, not just movie fans like me.

Take the freemium version of Watson Analytics in the Cloud for a spin here.

UPDATE: Now that the movie has been in American theaters for a couple of weeks, I decided to re-examine Twitter sentiment to see if the predicted post-release drop in negative sentiment actually took place.

This time, I use Watson Analytics to pull data from the hashtag #Minions from about a week before the release to about two weeks after its release.

Click to enlarge

Click to enlarge

 

I quickly create a Global Data Group in Refine (that I can reuse in other Watson Analytics features, such as Explore, Predict, and Assemble) for author following and a local data group in Explore based on days (2-9 July = pre-release and 10-21 July = post-release).

I then check the distribution of tweets across countries for the post-release period. Again, the United States leads the chatter.

Click to enlarge

Click to enlarge

 

I now take a look at the sentiments across days for only those authors with a high following (> 150 followers). As I predicted, the huge spike in positive chatter is owing to the release of Minions on July 10th.

Click to enlarge

Click to enlarge

 

Filtering this chatter for only on tweets from the United States, the steady decline of negative chatter after the film’s July 10th release is evident.

Click to enlarge

Click to enlarge

 

The data also reveal several other interesting trends. In the United States, for example, females consistently have led the Twitter charge on Minions.

Click to enlarge

Click to enlarge

 

In addition, after the release of the movie, females in the United States switched from sharing posts to writing more posts of their own.

Click to enlarge

Click to enlarge

 

Despite their overall excitement, females who did not like Minions expressed their feelings mainly through shares, and the majority of those negative sentiments came from Chicago, Illinois.

Click to enlarge

Click to enlarge

 

So, looking back on our original analysis of Twitter sentiment surrounding Minions, we noted that the greatest amount of negative sentiment about the movie pre-release was coming from the United States. However, the data also revealed that negative sentiment in other markets plunged after the release, indicating that a similar drop in negative sentiment might occur in the United States post release. And the post-release data reveal that the predicted drop in negative sentiment did indeed occur.

The ability to recognize this trend could have a profound impact across industries. For example, without this information, a large amount of negative sentiment preceding a product release could lead businesses to engage in an expensive, time-consuming and, ultimately, unnecessary marketing campaign. Or, worse, delaying or even abandoning the release. And for a product like Minions, which earned $115.2 million in its opening weekend in North America – the second-biggest opening for an animated film ever, this would have been a very costly mistake.

Suman Mukherjee works with the IBM Watson Analytics Product Experience and Design team. He creates demonstrations and works on customer POCs and internal enablements. He also loves to watch movies.

Related on Data Informed:

Chaotic Moon, IBM Watson Infer Audience Traits in Real Time

IBM’s Watson Eases Veterans’ Return from Military

Exec Tells How to Track and Understand Customer Sentiment

Use Data to Better Understand Customers and Pursue Prospects


Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise.


eBook: crack the unstructured data code with deep learning



Tags: , , , , , , ,

Post a Comment

Your email is never published nor shared. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>