Categorizing users of cloud services

Many services, e.g., in education and research, have witnessed increased productivity and scalability, mainly because of the growing prevalence of online platforms. To accelerate this progression, a detailed understanding of user interactions with these complex systems is instrumental. Current approaches for analyzing service user behavior have two main limitations: (a) unsupervised learning methods do not discriminate behavior meaningfully and scale poorly; and (b) surveys as input data probe only intentions. We introduce a framework to analyze user behavior in complex cloud services. Our objectives are (a) a computationally lean method to cope with large data sets and (b) for the input data to be free of assumptions. We define three data-driven metrics based only on the size and volume of data, using nested arrangements of zero and infinity norms. Using threshold analysis, user behavior data is categorized over one metric and subsequently verified over the other two metrics. We apply this method to analyze the behavior of users of nanoHUB services, the world's largest cyber-infrastructure for nanotechnology research and education. As input, actual user decisions are employed. The introduced frequency metric leads to a dispersion of usage duration. The resulting separated power and casual user categories are validated over the diversity and intensity metrics. Moreover, subcategories of previously unknown classroom users are identified. These findings allow policy makers to optimally tailor educational programs to such uncoordinated groups. Given nanoHUB's size, its leading role in educational services, and the approach's data-driven nature, the presented method is applicable to a wide range of cloud services.