An AI-based analysis of an exciting day of test cricket when India defended a target of 188 runs vs Australia.
AI Analysis of a test match
7th March 2017 will be a memorable day for test cricket as it witnessed one of the greatest battles between bat and ball. We thought this exciting day would be an ideal playground for our data science team to deploy our AI algorithms and get some insights about Indian fans’ behaviour on social media. We ran our analysis on ~46,000 tweets (excluding re-tweets) captured during the course of the day and grouped them into 10-minute slots to gauge how people’s emotions swayed as this nail-biting contest was playing out.
We assume the readers are aware of the ups and downs that were part of the last day, however for the uninitiated we recommend taking a look at the scoreboard on cricinfo’s website. India was looking good with two set batsmen (Pujara and Rahane) starting the 4th day and continuing their long partnership. The match started at 9:30 AM and Indian batting looked good for a first half hour. Then Aussies decided to take the new ball and their fast bowlers wreaked havoc on the bamboozled Indian batting side and took six wickets in a span of just 36 runs. The Indian fans got increasingly tensed and were aware that defending a 188 run target would be very difficult against a formidable opponent like Australia.
Indian bowlers were getting regular wickets but not at a pace that would be needed for defending such a tough target. When the Aussies scored their first 100 runs, they had lost only 4 wickets and looked poised to easily sail through to achieve the target. They say Test Cricket always finds a way to keep you hooked. We got a taste of this when Indian spinners (Ashwin and Jadeja) bowled one of the best spells of their time and got the remaining 6 wickets in a span of 11 runs. In a short span, India snatched this match from the jaws of defeat and our deep learning algorithms did a great job of capturing the shift in Indian cricket fans’ sentiment.
The sentiment was largely positive at the start as India was on a firm footing but it started to become increasingly negative when the batting side buckled. The sentiment continued to worsen as Aussies got closer to victory and then in a short span it was positive all over again as the Indian spinners were executing their turnaround performance.
As seen in our recent study on Oscars, our emotion analysis delves a level deeper and helps us understand the underlying emotions in play. Anger and Happiness were the most pronounced emotions and behaved the same way as sentiment.
We also analysed sad emotion tweets and noticed that sadness withered away as India’s chances of winning increased. The remaining sad emotion tweets could be attributed to some Australia supporters and some degree of noise in deep learning models.
Social Media + AI = Unparalleled Insights
At Karna-AI, we are pushing the boundaries on how AI can be used to get insights from news and social media. Our data science team is one of the thought leaders on this theme and are constantly working on new algorithms. We believe insights like the one we show in this blog, could be of great use for brands that conduct live events (sports teams, event agencies).
Want to improve your business in social media. Click here to schedule a free demo.