Data Mining from Process Models: Visualizations or Induction for Better Comprehension?
Tim Menzies, Siri‐on Setamanit, David Raffo · 2004
A standard methodology in the process simulation community is to build simulations using high-end visual programming (VP) system with powerful graphical front-ends. Many such VP tools exist including Vensim (see Figure 1), the Statemate state-based simulation model by i-Logix 1, the Extend discrete event simulation tool 2, just to name a few. Various benefits come from these VP tools including acceptance, debugging support and increased comprehension of the models. Acceptance: The visual appeal of these VP tools makes it easier to get user buy-in for the models. This can be especially important early in a fledging simulation project when developers need to achieve early results to gain the most interest from business users. Debugging: The tracing capabilities of these VP tools means that developers can fix erroneous parts of the models, sooner. Comprehension: Once the parts are fixed, the whole model is engaged and emergent trends of the entire system can be identified. These trends can be analyzed to make business decisions about (e.g.) better resource allocations at various points in times of the process being simulated. We take no issue with the first benefit since we have seen it all too often in our work with users. However, it is insightful to carefully review claims two and three. 1 Statemate and I-Logix are registered trademarks Ⓡ of I-Logix Inc. (3