Today’s datasets are increasingly growing in both size and complexity as the opportunities and techniques for data acquisition improve. This in contrast to the human cognitive capacity or perception (our ability to comprehend and process presented information), which stays essentially the same. The information overflow demonstrates the need for creating intuitive visualizations to display complex information, especially for decision making and data analysis.

The problem with common visualization techniques like bar and line charts are their limitation with respect to multidimensionality when for example visualizing multidimensionaldata from relational databases.

Based on previous research in the field of Information visualization and by prototype-driven development, a general framework called Stark for visualization of multidimensional data has been developed and evaluated in the context of scheduling statistics, which is strongly coupled to users and work shifts with starting- and ending-times. The work also included the development of a visualization tool, used for managing visualizations and configurations. The visualization tool was evaluated with respect to usability by applying the inspection method of cognitive walkthrough. The usability evaluation resulted in valuable knowledge of usability issues in the visualization tool, with respect to user feedback and ensuring expected behavior for actions.

The essential part of the framework consists of a largely customizable configuration, which is applied to a dataset in order to filter and group interesting data items. The configurations are built as trees of configurations, which defines how the datasets can be explored. This approach will enable visualizations for multiple stakeholder viewpoints.

The result indicates that the multidimensionality problem can be solved by defining levels of information, through the tree configurations. This will allow us to achieve a data partition, where the datasets are divided into custom layers, which can be explored by traversing the configuration. Then each level can be visualized using common visualization methods as bar and line charts, which humans are familiar with and can easily understand.