Published: 14^th , April 2026

Recursive Analytics for User Behaviour to Dynamic Stimuli Using a Rolling Window Based Precision

Yunus Fatima Abdulsalam, Fatima Isiaka and Lauri Matias Makela

Abstract

When recursive loop is coupled with a rolling window approach, where the most recent segment of behavioural data (for example, every 40seconds) is repeatedly re-examined while older observations gradually fall out of scope, the resulting pipeline achieves a level of temporal precision that is both fine-grained and adaptable to rapid shifts in context. The prediction accuracy (AUC ROC) here in this paper shows that the Baseline (static model) lies between 0.71μs and 0.03μs, a recursive model only between 0.78μs and 0.02μs, and a recursive/rolling window at 0.86s-0.01s performance. The mean accumulative error (MAE) in predicting how long a user would linger on a stimulus element dropped by 50% compared with the baseline, indicating a markedly finer grasp of temporal engagement patterns. This shows a user interface (UI) change only when predictive precision exceeds a predefined threshold. 

Keywords: Recursive Analytics, User Behaviour, Dynamic stimuli, Hyper-interactive digital ecosystems, Webpage content, Video content

ijaicp@aijournals .org

©ijaicp.aijournals.org  2019-2026