Applying Case Management Principles to Support Analytics Process Management

Analytics Process Management (APM) is an emerging branch of Business Process Management that is focused on supporting Business Analysts and others as they apply analytics approaches, algorithms, and outputs in order to discover and/or repeatedly produce business-relevant insights and apply them into on-going business operations. While APM is now occurring in many businesses, it is typically managed in ad hoc ways using a variety of different tools and practices. This paper proposes to use principles from Case Management (or equivalently, Business Artifacts) to provide a foundational structure for APM. In particular, six key classes of Case Types are identified, that can model the vast majority of activities and data being manipulated in APM contexts. These Case Types can simplify support for managing provenance, auditability, repeatability, and explanation of analytics results. The paper also identifies two key adaptations of the classical Case Management paradigm that are needed to support APM. The paper validates the proposed Case Types and adaptations by examining two recent systems built at IBM Research that support Business Analysists in the use of analytics tools.