Sentiment Analysis : Corona virus & lockdown impact

At the time of writing this, covid-19 cases globally have risen upwards of 1.4M cases and 82K deaths, with concerns that this pandemic will hugely overwhelm the hospital staff and its economic impact might lead us to recession or possibly even a depression.

However, this article is not about the prediction/ trajectory of covid-19 OR its economic impact.

As more and more countries declare a nationwide shutdown, most of the people are asked to stay at home and quarantined.

This article is about the sentiment analysis about the people of India and how they are feeling/coping up with this "lockdown”.

We have upwards of 5.3K cases in India as on 7th April and I hope that we would be successful in flattening the curve.

I have analyzed close to 90K+ tweets for 3 distinct periods during this lockdown and hope this data will give some insights:

• Day-0: 24th Mar - The day PM announced 21 days lockdown from 25th Mar until 14th April.
• Day-10: 3rd Apr – The day PM announcement to light up candles/diyas on 5th April as a sign to dispel the darkness spread by corona virus
• Day-12: 5th April – Post the candle light event at 9PM

Word cloud: The more often a specific word appears, the bigger & bolder it appears in the word cloud.

Sentiment Analysis:

• Day-0: We see people in general are neutral to positive about the lockdown (70%), perhaps indicating their commitment & support for flattening the curve.
• Day-10: The negative sentiments reduced by 4% with corresponding increase in positive sentiments.
• Day-12: There is further 7% reduction in negative sentiment immediately post the candle lighting event. This probably acted as morale booster and lifted the collective spirits during this lockdown.

However it would be an over simplistic approach to judge sentiments only in 3 polarities, as humans tend to exhibit multiple emotions at same time.

During every waking moment of life, a human mind consists of a variety of mental states. These mental states are typically named in commonsense terms as emotions, e.g., fear, disgust, joy, e.t.c.

Emotional Analysis — Gaining clarity on people’s emotion:

There are eight basic types of human emotions: Anger, Anticipation, Disgust, Fear, Joy, Sadness, Surprise & Trust.

We will do further text mining of each tweet, search for keywords which exhibit one or multiple of the above mentioned emotions and label them. Grouping all the words by emotions gives us following insights.

• The above plot shows that trust and anticipation are the top exhibited emotions and remain intact over this lockdown period.
• Display of disgust and fear have lowered over this period of time.

Lets hope we emerge successful in this fight against covid-19 and leave behind a healthier, happier and prosperous planet to the future generations.

Source: India covid-19 trend and maps courtesy of website: https://www.covid19india.org/

About the author- Shreyas is a seasoned telecom professional with focus on latest technology, cutting-edge analytics, producing actionable insights and loves to work on data science projects. He has more than 19 years of experience in network solutions, technical consulting, analytics and business development with specializing across various domains and has worked on multi-cultural diversified geographies like Asia, Middle-east and North America.