Visual-interactive exploration of interesting multivariate relations in mixed research data sets

摘要: The analysis of research data plays a key role in data-driven areas science. Varieties mixed sets exist and scientists aim to derive or validate hypotheses find undiscovered knowledge. Many techniques identify relations an entire dataset only. This may level the characteristic behavior different subgroups data. Like automatic subspace clustering, we at identifying interesting attribute sets. We present visual-interactive system that supports explore between aggregated bins multivariate attributes abstraction enables application statistical dependency tests as measure interestingness. An overview matrix view shows all attributes, ranked with respect interestingness bins. Complementary, node-link reveals bin by positioning dependent close each other. information drill-down based on both expert knowledge algorithmic support. Finally, subset clustering assigns groups. A list-based cluster result representation scientist communicate findings glance. demonstrate applicability two case studies from earth observation domain prostate cancer domain. In cases, enabled us most relations, already published results, and, moreover, discover unexpected relations.