Collaborators: Data Science and Contextual Marketing teams at Globys, Inc.
|
We are applying machine learning at big data-scale to automatically detect, analyze and charaterize the behavior of millions of mobile phone subscribers. The way mobile phone subscribers use their phones can be captured by "usage fingerprints" computed from the carriers' billing data. Similar fingerprints form cohesive behavioral profiles that help marketers devise highly targeted, personalized offers aimed at increasing subscriber loyalty and carriers' revenues. Our approach to behavioral fingerprinting differs from standard business intelligence techniques in that we capture and act upon the dynamic behavior of each subscriber as it evolves over time along different dimensions of usage.
We address the challenges posed by the dynamic nature and huge volume of the subscriber data by scaling up state-of-the-art machine learning algorithms to big data-regimes, while hiding the complexity behind easy-to-use interfaces that enable non-technical marketing experts to determine whom, when, how to message with high precision.
|