Interactive Poster: Interactive Visualization approaches to the Analysis of System Identification Data

a,b) One region with large residuals c,d) Two regions with large residuals Figure 1: A semi-immersive display showing the relationship between the model input, the residual (the error between the model and the real system) over time. The visualization reveals that in some regions the residual is quite large, i.e. the model does not fit very well. The temporal information is crucial for the model validation which is an important step in the system identification process. We propose an interactive visualization approach to finding a mathematical model for a real world process, commonly known in the field of control theory as system identification. The use of interactive visualization techniques provides the modeller with instant visual feedback which facilitates the model validation process. When working interactively with such large data sets, as are common in system identification, methods to handle this data efficiently are required. We are developing approaches based on data streaming to meet this need