Abstract:

This paper is framed within a competitive funded education innovation project -PIME/25-26/574 Early Signals: Predictive Analytics and Proactive Tutoring to Reduce Dropout in Continuous Assessment Courses- that focuses on reducing early withdrawal in Introducción a la Arquitectura, a high-enrolment, first-year course delivered over a single semester (4.5 ECTS). The project, developed at the School of Architecture of the Universitat Politècnica de València, investigates the potential of a data-informed, early-warning model to mitigate early withdrawal and strengthen both academic performance and student satisfaction in a continuous-assessment context. The target cohort reaches up to 360 students. The primary objective of the project is to integrate learning analytics with proactive tutoring so as to detect early indicators of potential student dropout and intervene at an early stage. However it also aims to reduce early withdrawal to ≤15%, to raise average achievement by ≥20%, and increase student satisfaction by ≥20%.

In order to do so, an Early Warning Panel (EWP) connected to Sakai-PoliformaT, the institutional learning management system of the Universitat Politècnica de València, is designed. EWP integrates the data extracted from processing weekly traces (logins, on-time/late submissions, early quiz/project grades) and produces risk estimates through a lightweight logistic regression calibrated on historical data (2014–2024). All data flows follow ethical approval, pseudonymisation, and role-based access controls.

This article presents the context, methodology, objectives, and preliminary first-year hypotheses. At this stage, the preliminary results from the baseline analysis appear to indicate a clustering of withdrawals among students with low platform activity in weeks 1–3 and a missed first milestone around week 5, validating the project’s preventive focus and informing initial thresholds for risk detection.

Aligning early-warning analytics with proactive and standardised tutoring emerges as a feasible and transferable strategy for first-year, single-semester courses in Architecture. The value of the project lies in the development of a transparent and replicable workflow that defines data-extraction procedures, relevant indicators, threshold criteria and corresponding actions, documented as a reusable institutional kit. It also lies in a tutoring protocol that strengthens student persistence, academic achievement, and satisfaction while preserving pedagogical autonomy.

Keywords:

Early warning, learning analytics, proactive tutoring, first-year retention, Sakai-PoliformaT, continuous assessment.