Data Mining Challenges in Automated Prompting Systems

With the rising cost of medical treatment and majority of the aging population preferring an independent lifestyle, the need for assistive technologies and smarted devices are increasing like never before. A prompting system is a technique that provides interventions to a smart home inhabitant in order to ensure successful completion of an activity. Machine learning techniques could be used to automate this system but it comes with the challenge of imbalanced class distribution naturally occurring in the data. In this paper, we comparatively analyze two techniques (sampling and cost sensitive learning) to deal with this challenge. Author