http://www.digitaljournal.com/tech-and-science/technology/mining-3-5-billion-tweets-confirms-sunny-weather-makes-us-happier/article/503583

Mining 3.5 billion tweets confirms sunny weather makes us happier

Posted Sep 27, 2017 by James Walker
The largest study of its kind has demonstrated that good weather makes people feel happier by analysing over three billion Facebook and Twitter posts made over a seven-year period. The volume of data assessed for the study far exceeds previous attempts.
3.5 billion tweets have confirmed sunshine creates happiness
3.5 billion tweets have confirmed sunshine creates happiness
Pexels / Julian Jagtenberg
The work was completed by a team of Stanford University researchers led by Patrick Baylis. As the MIT Technology Review reports, the study is based on the premise that the tone of our social media posts tends to reflect our overall mood.
The team mined over 3.5 billion Facebook and Twitter posts made between 2009 and 2016. The status updates and tweets all originate from one of 75 U.S. metropolitan areas. The volume of data collected suggests the findings should apply in other regions across the world.
Using AI-powered sentiment analysis technology, the researchers processed each post to determine whether it expressed a positive or negative mood. The result was then compared to the weather conditions in the area at the time the post was published. By looking at the location of the post, accurate weather data could be obtained for the exact date and place it originated from.
Across the aggregate data set, the overall findings are conclusive. People post "happier" status updates when the weather outside is an ideal blend of warmth, cloud cover and humidity. Posts made when the weather is cloudy, wet or uncomfortably humid tend towards a negative sentiment expression, confirming human mood can be significantly altered by meteorological conditions.
READ NEXT: 'Critical need' for IoT experts who can mine digital data
Baylis' team isn't the first to report this finding. The study is the "largest ever" investigation of its type though, considerably exceeding other attempts. Previous research efforts have also tended to focus on social media posts but have been limited to smaller datasets. A 2013 report from the University of Vermont looked at 37 million messages while a 2014 Stanford paper was based on two years' worth of Twitter data from 32 U.S. cities.
Although the result might not seem to have practical implications, businesses could use the information to optimise their social media campaigns for maximum real-time effect. A firm could find a post is more engaging if it directly references the weather or tries to make light of a rainy day.
The discovery could also impact artificial intelligence and digital assistants. An AI could use the weather as a signal of your mood, giving it more insight into how you're likely to behave.
There's also the possibility that weather-based ad targeting could become a reality. On bad weather days, ad platforms could show you ads for products designed to make you feel happier. It would businesses a way to cash in on rain or cloud by selling you digital media or tempting you to an indoors event.