Footnote:
Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments September 1, 2020 erstellt
Description:
We propose an interpretable deep survival model that can capture human-understandable nonlinear patterns from big data while handling censored observations and time-varying customer dynamics. To this end, we build WTTE-TCN (Weibull Time to Event Temporal Convolutional Networks) and apply post-hoc eXplainable Artificial Intelligent (XAI) methods to explain model predictions in a human-interpretable manner. When applied to mobile game churning data, WTTE-TCN demonstrates superior performance with less computational costs while also addressing the limitations of traditional survival models such as time-varying covariates.We build the algorithm so that managers can easily interpret human-understandable explanations and draw actionable insights and inform potential prescriptive strategies. For example, replaying the cleared stages in the game is linked to early churning, whereas making users feel a higher sense of accomplishment through appropriate hardness is connected to decreased churning. We identify misdesigned in-game systems (e.g., difficulties and social interaction mechanisms) that increase churn rate and provide suggestions to improve them