Prior research suggests that weather conditions may substantively impact people's emotional state and mood. In Russia, the relationship between weather and mood has been studied for certain regions-usually with severe or extreme climatic and weather conditions-but with quite limited samples of up to 1,000 people. Over the past decade, partly due to the proliferation of online social networks and the development of natural language processing techniques, the relationship between weather and mood has become possible to study based on the sentiment expressed by individuals. One of the key advantages of such studies based on digital traces is that it is possible to analyze much larger samples of people in comparison with traditional survey-based studies. In this article, we investigate the relationship between historical weather conditions and sentiment expressed in seven Russian cities based on the data of one of the largest Russian social networks, Odnoklassniki. We constructed a daily city-level expressed positive sentiment metric based on 2.76 million posts published by 1.31 million unique users from Odnoklassniki and studied its dynamics relative to daily weather conditions via regression modelling. It was found that a maximum daily temperature between +20 °C and +25 °C, light breeze (between 5 and 11 km/h) and an increase in the average daily temperature by 20-25 °C compared to the previous day are all associated with higher numbers of expressions of positive sentiment, whereas the difference between the maximum and minimum daily temperatures of 15-20 °C is associated with lower numbers of expressions of positive sentiment.
Keywords: Natural language processing; Russia; Sentiment; Sentiment analysis; Weather.
© 2022 Smetanin.