Designing a Predictive Analytics System for Student Academic Counselling at Federal Polytechnics

Federal polytechnics in Nigeria train large numbers of technical students whose academic trajectories could benefit significantly from early data-driven counselling interventions. This study designed a predictive analytics system for supporting academic counselling at three federal polytechnics: Federal Polytechnic Bida (NC), Federal Polytechnic Ado-Ekiti (SW), and Federal Polytechnic Nekede (SE). A professional system design methodology was applied, with structured consultations with 15 academic counsellors and 10 polytechnic IT managers, assessment of available student data systems, and benchmarking against early alert systems deployed at South African universities and Purdue University's Course Signals system. Available data sources included ND1 and ND2 continuous assessment scores, attendance records, financial aid status, and course withdrawal history. The system designed specifies a weekly risk score update engine using logistic regression trained on historical outcome data, a counsellor alert dashboard with personalised intervention recommendation text, student-facing self-service performance tracker, and data privacy controls compliant with NDPA. Minimum data infrastructure requirements for each polytechnic are specified with cost estimates. Expert review by nine educational data science and academic advising specialists confirmed the system's pedagogical appropriateness and technical feasibility. The study recommends starting with the Bida pilot where student data quality was highest, and establishing an inter-polytechnic data science community of practice to share model improvements.

Keywords: student early alert system, academic counselling, federal polytechnics, predictive analytics, educational data science