Dataming approaches to Quality of service performance in Nigerian GSM Networks

Abstract

Quality of Service (QoS) disparities persist in Nigerian GSM networks despite regulatory efforts. This paper presents a data-driving analysis of QoS performance using 2,944 real-world call records collected via the KayusQoS mobile framework from M, A, G, and E in Nasarawa State. Five Key Performance Indicators (KPIs): Call Setup Success Rate (CSSR), Call Drop Rate (CDR), Handover Success Rate (HOSR), Received Signal Strength (RSS), and Retainability Rate (RetR) were aggregated into a composite QoS score. Pearson correlation identified a moderately strong positive relationship only between M and A (r = 0.531, p < 0.001). Analysis of Variance (ANOVA) revealed no significant location differences (p = 0.860) but highly significant operator differences (p < 0.001), with M achieving the highest mean composite QoS (4.02). Multiple regression provided excellent predictive accuracy (R² = 0.915). Findings underscore operator-specific infrastructure gaps and demonstrate the effectiveness of data mining approaches for actionable QoS insights.