Abstract for: A Qualitative Approach for Clustering Simulation Output Data Based on Behaviour Mode Analysis

Techniques such as regression, classification, and clustering methods have been applied in sensitivity and behavioural analysis to explore the impact of policy changes. Many of these techniques include input from the user (number of segments, weights) etc., providing the user more control in finding different results based on their clustering needs. One problem which occurs repeatedly in the literature, is finding ways to easily deal with the clustering of oscillations, and features including the trend and amplitude often need to be included in models in order to deal with them effectively to use extra features. Here we propose a method based recent research that we find clusters oscillations quite well, without the need to use extra features, only possible ranges of important parameters.