Almost 60% of service providers in this month's Thought Leadership Council (TLC) say machine learning will become a critical part of their company's network operations by 2020. Not surprisingly, perhaps, the majority of those service providers will use machine learning first in predictive maintenance.
In fact, 91% of panelists in the Council said machine learning is considered a critical technology in predictive maintenance. One provider explained that machine learning "is easily integrated into predictive maintenance because it takes advantage of automated pattern recognition and can be used to address [an] operator's biggest pain point: service-level degradation, which in turn helps prevent customer complaints and reduces churn."
Indeed, the message from service providers in this month's forum was clear: they believe machine learning can be a valuable asset, and they're finding small but interesting ways to apply it right now. For machine learning to gain more ground, however, developers have their work cut out, as revealed in the latest TLC report, Emerging Tech Focus Group: Operators -- Prove Machine Learning's Applicability. Council members were asked a series of nine questions about machine learning to ascertain current and future plans, how well it's understood and influences impacting the technology.
About half the panelists in this month's forum say their companies already have deployed machine learning, and several discussed ways in which it's already being deployed. One panelist explained how machine learning is being used by his company to boost sales in the international B2B market, where his company has a low win ratio due to several factors, including aggressive market competition, a variety of competitor products and a variety of customer needs in requests for proposal (RFP).
"Machine learning was used to capture the hidden patterns of lost and won bids and score the bids according to a classification algorithm," he said. It has been so successful that the company is now utilizing algorithm recommendations to determine the best way to sell standard products in the global market for MPLS, voice wholesale minutes, Internet and other similar products.
Despite these positive signs for machine learning's applicability, panelists were quick to point out that machine learning isn't always easy to understand or apply. More than half of panelists say the biggest obstacle machine learning faces is a lack of clear understanding in terms of how it can be applied.
As one panelist said, "There isn't a clear and consolidated experience regarding the concrete results that machine learning can provide."
TLC is a Heavy Reading research initiative that consists of panels of CSP experts focused on key areas of telecom development, including service assurance, SD-WAN, automation, 5G and IoT. Members participate in Q&A forums several times a year, and all information disclosed by Council members remains anonymous.
— Denise Culver, Director of Online Research, Heavy Reading