Externalization of Data Analytics Models: - Toward Human-Centered Visual Analytics.

Visual analytics tools (VATs) can support the execution of complex cognitive activities. Most VATs make use of analytics models in the execution of data-intensive activities. However, due to their non-transparent and black box nature, analytics models are hard to use. This leads to a lower degree of understandability of the models, which then results in a lack of trust and applicability. To overcome this problem and to help bring VATs closer in line with users' mental models, we introduce a framework for the externalization of analytics models in VATs. In doing so, we propose the use of scaffolding to provide the users with visual support in the execution of their cognitive activities. The aim is to make VATs more human-centered. To demonstrate the application of the externalization framework, we present two components of a visual analytics tool named VARSITY.