Humidity and air temperature predict post count on Twitter in 10 countries: Weather changes & LIWC psychological categories

Abstract

There are many indications that weather conditions influence human life and well-being. Some of these indicators, such as the influence of weather on human health, have been explored in detail. On the other hand, the influence of daily fluctuations of different meteorological variables on the human psychological state still remains unknown. We apply combined methods from statistics, psychology, machine learning, and complex networks theory to explore the influence of weather parameters on different psychological categories of Twitter users in ten different countries. Our results show that the temperature, pressure, and humidity are highly correlated with Twitter users' activity, sense, and affect. Our comparative analysis for different countries shows that the strongest correlation was found for the USA, Italy, and Portugal, indicating differences between countries. However, our results show that the level of activity of Twitter users, described as Post Count, is strongly connected to changes in temperature and humidity in all countries. We use complex networks theory to explore these connections and differences between countries further. Our findings suggest that weather parameters can be used to predict Twitter users' activity and psychological manifestations, which can be beneficial to marketing and advertising.