Data Prepared
Gathered, cleaned, and structured network data.
A massive dataset of 275+ million records
Completed extraction, transformation, and cleaning of all necessary performance, cost, and utilization data for ML modeling.
Correlation Analysis
Identified meaningful patterns and dependencies.
26 different features used
Performed correlation analysis to identify relationships between investment, utilization, and network KPIs.
ML Prediction Ready
Trained and validated model for network performance.
LightGBM • Accuracy: 92%
Selected and trained ML models to forecast network performance trends and potential degradation.
Results
Optimization potential & investment opportunities.
80.41% Opt-performing • 10.57% Over-performing • 9.02% Under-performing
Model-driven segmentation highlights where to optimize versus where to invest.
