Improving service for Charter with machine learning
Charter Communications is one of the largest telecommunications companies in the US and is committed to offering integrated, cost-saving services and quality customer service. The company has traditionally provided cable and broadband and is now expanding to offer a cellular service product, Spectrum Mobile. Since Charter does not have the infrastructure to offer cellular service to its customers, it turned to PK to build a solution that will help customers’ devices connect in the most efficient way.
Charter’s cellular service offering, Spectrum Mobile, is a mobile virtual network operator (MVNO). As an MVNO, it doesn’t own cellular spectrum and must lease seconds of service from a mobile network operator (MNO). This can be expensive, so to limit the seconds of service leased, the MVNO makes sure that devices automatically connect to Wi-Fi as much as possible, while still providing the most reliable service possible to customers.
PK’s machine learning solution categorizes Wi-Fi networks based on reliability and performance, creating a ranked prioritization based on the best performing network to least. This reduces costs by eliminating unnecessary cell tower usage by devices. The solution also predicts where new networks will rank on the prioritization via phone event data.
In the future, this solution can provide more advanced ways to approach problems through reinforcement learning, where an algorithm runs natively on devices and makes adjustments through what it has learned from the device. Other potential uses of this technology include identifying when a customer’s current network will drop so that the device can switch networks before the outage happens. PK is currently delivering scalable, repeatable data and machine learning modeling pipelines in order to act on these opportunities.