Mining of productive periodic-frequent patterns for IoT data analytics

摘要

Healthcare applications in Internet of Things (IoT) systems have been increasingly researched because they facilitate remote monitoring of patients. Though IoT may create data consisting of much useful information, finding meaningful patterns in huge amounts of IoT data is a challenge. In this paper, we propose a new type of behavioral pattern called productive periodic-frequent sensor patterns (PPFSP). PPFSP can find a correlation among a set of temporally frequent sensors patterns which can reveal interesting knowledge from the monitored data. We also present two approaches to discover PPFSP; a parallel method using a compact productive pattern sensor tree (PPSD-Tree) and Map-reduced PPFSP-H mining algorithm on Hadoop to facilitate PPFSP mining on large data. Results show that our methods are both more time and memory efficient in finding PPFSP than the existing algorithms.