CHEERUP: A General Software-Environment for Building, Using and Administering Predictive Monitoring Portals

Abstract

The intended meaning of the term predictive monitoring used in the paper is the following. A population of subjects (living beings, machines, works of art, etc.) is monitored by a domain expert with regard to the possible occurrence of an undesired/desired event E. More precisely, an expert periodically (e.g. every two years, every week, etc. depending on the specific application) examines the single subjects and, for each of them, enters examination outcomes in a database where statistical data are automatically processed in order to produce probabilistic inferences about the occurrence in the future of E for the subject under examination (individualized prediction). This allows the expert to take suitable measures in advance in order to prevent/favour the occurrence of E for the subject. Such an approach to predictive monitoring requires that the expert who monitors subjects has at his/her disposal a suitable software system provided with database and algorithms for both properly managing monitoring-processes and producing probabilistic predictions. The paper presents CHEERUP : a prototype product, usable via Internet, that consists in a general software-environment for building, using and administering specific predictive monitoring software-systems (in the paper called portals).

Keywords

• computer applications
• predictive models
• learning systems
• data processing