Visual Causality Analysis of Event Sequence Data

Causality is crucial to understanding the mechanisms behind complex systems and making decisions that lead intended outcomes. Event sequence data widely collected from many real-world processes, such as electronic health records, web clickstreams, financial transactions, which transmit a great deal of information reflecting causal relations among event types. Unfortunately, recovering causalities observational sequences challenging, heterogeneous high-dimensional variables are often connected rather underlying excitation hard infer limited observations. Many existing automated analysis techniques suffer poor explainability fail include an adequate amount human knowledge. In this paper, we introduce visual analytics method for in data. We extend Granger causality algorithm on Hawkes processes incorporate user feedback into model refinement. The visualization system includes interactive framework supports bottom-up exploration, iterative verification refinement, comparison through set novel visualizations interactions. report two forms evaluation: quantitative evaluation improvements resulting user-feedback mechanism, qualitative case studies different application domains demonstrate usefulness system.