Python

In this thesis a software for model behavior testing, validation and analysis is created. This software integrates previously developed methods on steady state behavior analysis based on traditional statistical methods and behavior classication based on pattern recognition literature, and introduces new features, functionalities and usage modes to them. It is the rst comprehensive tool that communicates with existing modeling software and performs automated analysis and evaluation. It is shown that this tool can assist the modeler/analyst in various stages during a dynamic modeling study.

The project involves developing and testing an algorithm for clustering oscillatory time-series data that is generated by dynamic simulation models. The team is expected to review alternative approaches from data mining and pattern recognition fields, and apply them on a sample dataset in order to compare and contrast the performance of alternative approaches.