Research On Attendance Prediction In Social Networks Considering The Weather's Influence

Understanding the issue of people participating in real-world activities has always been a subject of active research and can provide valuable insights into human behavior analysis and prediction / advertising related to real-world activities.The emergence of a new type of social network,Event-based social network(EBSN),has attracted more users on the network to organize or participate in an offline event,thereby stimulating new research in this field.When an event organizer plans an event,it needs to have a certain degree of certainty about the user's attendance at the event.However,most of the existing research focuses on the analysis and extraction of the factors of the activity itself(such as the theme of the activity,the distance of the activity,etc.),and few studies have analyzed the external factors of the activity,such as weather factors.At the same time,the existing work is not very accurate in predicting certain activities(such as outdoor activities).Therefore,in the work of this article,we use the large-scale event data sets collected from the wellknown event social network Meetup in two cities,London and New York,to study the potential factors that affect user attendance.This article innovatively uses activity duration and weather factors as the two key factors that affect the user's attendance behavior,and fully understands and explores the different ways that weather factors affect different types of activities,and then combines the gradient tree-based lifting tree algorithm to The user's behavior at the event was predicted.Experiments show that due to the addition of weather factors to the prediction model,the prediction model in this paper achieves better prediction accuracy.The main contributions of this article are summarized as follows:1.This paper conducts an in-depth study of weather affecting users' attendance.First,extract and classify the temperature,humidity,wind direction and other characteristics of the weather factors to make a foundation for exploring the impact of weather on different types of activities;at the same time,we divide fine-grained weather factors into direct and indirect effects.2.This paper identifies some key features that can affect the user 's attendance at the event and extracts them comprehensively.This includes user interest features,user activity distance features,activity duration features,and weather features.3.This paper integrates a variety of factors of the activity and combines the gradient descent-based lifting tree to design an event attendance prediction model that combines weather,namely GBT-W.In the modeling process,I deeply explored the selection of negative samples.4.This paper conducted a comprehensive test of the designed model,and the experiment showed that the weather factor is effective in improving the accuracy of the prediction model.At the same time,based on this prediction model,we considered the influence of the uncertain weather conditions in the real environment on the model and introduced the specific use method of the article model for the event organizer.