Survival analysis for user disengagement prediction: question-and-answering communities' case

TL;DR

This study applies survival analysis to model and predict user disengagement in three Stack Exchange communities, revealing that active users and those with positive content views tend to stay engaged longer across different topics.

Contribution

The paper introduces the use of survival analysis techniques to predict user disengagement in Q&A communities, demonstrating consistent patterns across diverse communities.

Findings

Active users are more likely to stay engaged longer.

Positive content perception correlates with prolonged engagement.

Patterns hold across different community themes.

Abstract

We used survival analysis to model user disengagement in three distinct questions-and-answering communities in this work. We used the complete historical data of {Politics, Data Science, Computer Science} Stack Exchange communities from their inception until May 2021, which include the information about all users who were members of one of these three communities. Furthermore, formulating the user disengagement prediction as a survival analysis task, we utilised two survival analysis techniques to model and predict the probabilities of members of each community becoming disengaged. Our main finding is that the likelihood of users with even a few contributions staying active is noticeably higher than the users who were making no contributions; this distinction may widen as time passes. Moreover, the results of our experiments indicate that users with more favourable views towards the content shared on the platform may stay engaged longer. Finally, the observed pattern holds for all three communities, regardless of their themes.