By Jason Rupe & Colin Justis, CableLabs
Maintenance is necessary, but service disruption isn’t. Some cable system repairs will impact service in ways that customers notice, but can be necessary and urgent. Fixing service while a customer is not using services is far better. But to do that without bothering the customer requires a usage forecast model.
With historical usage data of service classes, we created a simple model that predicts the amount of data being consumed by end devices in a cable plant. Using this model as an indicator of usage by customers, we can determine the best time for a repair interval of a defined duration, allowing a technician to time necessary but disruptive operations to minimize the disruption to customers. We analyzed the data, tested the model, then built a simple application based on the model which can specify the best time(s) to conduct a service disruptive repair for a defined duration, for a given set of end devices to be impacted.
The application is about to be tested in a trial. Customer call-in rate will be used to measure the effectiveness of the projected schedule, as compared to a baseline.