Hello Dear Reader,
Back in November we revisited out series on Twitter Sentiment Analysis with a review of the Thor: Ragnarok sentiment and a prediction for the Justice League movie. The prediction for the Justice League movie was off. We were not alone. The film missed even modest industry estimates and under performed. When we made our initial estimate of $118.5 million we were only $500,000 off the estimate of Box Office Mojo's $118 million. Instead of confirmation we had to rethink our assumptions. As with anything in the data science process if you get unexpected results, it raises new questions.
The movies we'd captured thus far had all had positive sentiment. These where negative. What did that mean and how would it affect the box office? We had never collected sentiment before a movie came out. What is the normal sentiment pattern and could that have indicated box office performance?
In the cases we have examined thus far, we noticed that sentiment taken before a movie premiers is higher than that of when it premiers. The sentiment before the premier was 57%. Where we went wrong was assuming this correlated with what we should expect opening weekend. Once the movie opened it continued to drop.
Thursday's positive sentiment was only 47%.
Friday was 45%.
Saturday was 44%.
Sunday we stopped the collection half way through the day and it was at 51%.
The high for the entire weekend was 51%, this was 6 points off the sentiment taken on Monday and for the weekend was only at 46%. This was 11 percentage points off the high sentiment. In the case of Justice League this showed in a lower box office. Justice League was the first movie this year we tracked that did not make back its budget in Domestic Box Office gross.
STAR WARS: THE WORST SENTIMENT
Then came Star Wars. Star Wars has made back it's budget. The real story here is how far off the previous Star Wars movie in the same trilogy it has fallen. The box office is not yet closed, but as a year over year average Star Wars: The Last Jedi is behind Star Wars: The Force Awakens by a staggering $242.8 million in Domestic Box Office gross. Star Wars: The Last Jedi has the lowest positive sentiment of any movie tracked so far.
As you can see from the visual even though Star Wars has made it's money back by over $300 million, its positive sentiment was 33%. That is staggering.
This has lead us to look at additional sources. You'll notice the IMDB and Metascore ratings are on each report.
SO WHAT NOW
New questions are awesome. What we've decided is we need more data, we need to branch out to different movie genres and see if patterns hold up. To do this we need to be able to collect data from Twitter historically, we wanted to add additional scores like IMDB and Metascore. We want to see what patterns emerged.
Is sentiment always lower after a movie premiers vs. the sentiment before? What does that drop mean? If you have a strong brand, like Star Wars, with low sentiment can you still be profitable at the box office? If you have low sentiment, an established brand, and the movie is not profitable is that indicated in social media sentiment?
To answer these questions we will use python to scrape and collect data. Then we will review those patterns and see what we can find. In our next blogs we will review how we scraped the data, ingested it to SQL DB, and used Azure ML for semantic scoring.
As always Thank you for stopping by.
This post first appeared on MSDN Blogs | Get The Latest Information, Insights, Announcements, And News From Microsoft Experts And Developers In The MSDN Blogs., please read the originial post: here