Our Solution
Luma Analytics developed a machine learning model trained on three years of historical hourly transaction data and branch demand patterns. Using decomposition and temporal pattern isolation algorithms, the model identifies cyclical and seasonal trends to forecast expected transaction volumes. The fully automated, self-evaluating solution ingests new data and retrains monthly – ensuring accuracy and adaptability over time.
Outcomes
The model provides accurate two-month forecasts of branch transaction volumes and full-time equivalent (FTE) requirements at an hourly level. This has enabled the bank to optimise rostering and scheduling decisions, reduce manual forecasting effort, and improve service levels across its retail branch network. Automation has also delivered ongoing efficiency gains and laid the groundwork for future AI-driven workforce planning initiatives.
Featured Results
- Machine learning model predicting branch demand and staffing needs
- Automated monthly retraining for continuous accuracy and performance
- Two-month, hourly-level forecasts improving workforce planning
- Reduced manual forecasting effort and improved operational efficiency
- Enhanced customer experience through optimised branch resourcing
