Hokuyo Bank developed an AI-powered congestion prediction model to forecast congestion at branches and ATMs, and began displaying the AI Congestion Prediction Calendar at branches on 1 October 2026. The initiative is intended to reduce visitors' waiting times and improve branch operating efficiency.
Hokuyo Bank developed an AI-powered congestion prediction model to forecast congestion at branches and in-branch ATMs. Replacing the conventional common calendar, the model provides branch-specific forecasts using AI, enabling visitors to choose less congested times. The model's forecasts are provided in calendar format.
The model uses several years of past queue-ticket issuance records, weather information and other data. In addition to basic trends such as the day of the week and time of day, it learns special factors such as the date and payday to predict daily congestion at each branch. The AUC, a machine-learning evaluation metric, was 0.89.
By sharing the AI Congestion Prediction Calendar with customers, visitors can choose times that avoid congestion, which is expected to reduce waiting times at counters. The bank is also advancing the optimisation of staffing allocations, including adjusting employee shifts on busy days, while seeking to improve operating efficiency on quiet days by consolidating routine operations.
Hokuyo Bank said it will continue to improve customer services and branch operating efficiency. The source of this news is a Hokuyo Bank news release.