Return to the Ideas Board: Project and Papers
6/7
Title:
How weather changes are connected to expressiveness on Twitter
Co-autors:
Ljubisa Bojic, Ljiljana Lazarevic and [add co-author here]
Target journals:
CyberPsychology & Behavior
Abstract Draft:
Posts from more than 111.000 Twitter profiles from 10 countries have been analyzed in terms of selected linguistic categories and post count each day for six months and correlated with live weather data from these locations including pressure, humidity, wind speed, wind direction, cloudiness, rain volume, snow volume, and air temperature. We found 431 correlations between weather and measured parameters, out of which 9 were strong, 170 moderate and 252 weak. Temperature (123), pressure (80) and humidity (70) were weather parameters with the most correlations with measured categories. On the other hand, locations that had most correlations with weather changes were in US (68), Portugal (58) and Italy (54). Measured categories that had highest number of correlations with weather changes were Post Count (49), Senses (26) and Affect (24). Post Count was strongly connected to changes in temperature and humidity in all countries. Thus, there are indications weather changes are connected to overall expressiveness on Twitter. This knowledge can be used to predict Post Count and expressiveness by using weather forecast, which can be applied to different spheres of life, including marketing. Further research may include predictive AI algorithms.
Data file with correlations can be downloaded here.