Rain sends us all a-Twitter with emotion

The March rain sucks, right, but it’s better than the February heat wave, no?
成都夜生活

Yes and no, says one of the world’s first big data tools to monitor collective emotions on a massive scale.

How we all feel at any given time is hard to know. Surveys suffer from relying on people to remember how they felt at a particular time in the past.

We Feel is a data tool from the CSIRO that monitors population emotions in real time by analysing words used in Twitter posts, on a large scale and in real time. It shows Sydneysiders expressed more joy over three days of rain this week than they did during the February heat wave.

Joyful tweets made up an average of 15.42 per cent of English language tweets from Sydney on the three wet days from Wednesday to Friday, which was more than the 11.47 per cent of joyful tweets during three days of heat from February 9-11, when the thermometer soared to 41 degrees.

And if Twitter users are representative, the heat made us crankier too, with anger coming out in 1.15 per cent of tweets in the heat, compared with 1.07 per cent during the rain days.

The differences between February and March may have been exaggerated by system maintenance but the overall patterns seem clear. The tweet stream also seemed to confirm the rainy day blues, registering an average of 4.12 per cent sad words over the three wet days compared with 3.16 per cent over the February scorchers.

“We wanted to see if we can monitor people’s emotional states in real time over parts of the country” says Cecile Paris, senior principal researcher at CSIRO’s Data61.

“We hope it can uncover, for example, where people are most at risk of depression and how the mood and emotions of an area or region fluctuate over time,” said Dr Paris. “It could also help understand questions such as how strongly our emotions depend on social, economic and environmental factors such as the weather, time of day, day of the week, news of a major disaster or a downturn in the economy,” she said.

Project collaborator The Black Dog Institute is using the tool to better understand the prevalence and drivers of emotions. The tool processes 45,000 Twitter posts per minute, on average. The results are visualised online in real time and freely available for use.

Joy has been dominant emotion every day so far this year with peaks on New Year’s Day and most recently, International Women’s Day on March 8. Women tweeted more joy than men in and New Zealand that day, most frequently using words such as happy, great, good, proud, strong, smart and successful. They also tweeted more joyfully than their female counterparts worldwide, with 20.09 per cent of their tweets expressing joy compared with 14.45 per cent the world over.

So far the data has confirmed typical known patterns of mood through the day. Peaks of sadness have been observed in relation to events such as the deaths of cricketer Phillip Hughes and comedian Robin Williams. But “joy tends to be quite high most of the time. It is usually predominant,” says Dr Paris.

“With major events we see sadness increasing a lot, but it doesn’t necessarily overcome the joy either.”

Tweeters expressed the most sadness so far this year on January 20, the day of Donald Trump’s inauguration. An anger peak was reached nine days later, coinciding with widespread dismay and demonstrations at US airports in response to the first ban on immigration to the US from seven Muslim countries.