Changes between Version 8 and Version 9 of AdditionalMaterial/ECMLPKDD
 Timestamp:
 06/28/12 13:03:37 (12 years ago)
Legend:
 Unmodified
 Added
 Removed
 Modified

AdditionalMaterial/ECMLPKDD
v8 v9 15 15 16 16 == Case Study: Tower Data == 17 The demonstration will show how HeuristicLab can be used for knowledge discovery in a real world application, in particular, for finding relevant driving factors in a chemical process and for the identification of whitebox regression models for the process. We use the [[ tower data set]] which is kindly provided by Dr. Arthur Kordon from Dow Chemical (also see http://www.symbolicregression.com/?q=towerProblem). First we use symbolic regression to create a whitebox regression model for the chemical process. Afterwards we describe how the algorithm can be used to determine the most important driving factors for the process. In the end we show how symbolic regression models can be manually simplified and how visual hints can guide the user in the simplification process.17 The demonstration will show how HeuristicLab can be used for knowledge discovery in a real world application, in particular, for finding relevant driving factors in a chemical process and for the identification of whitebox regression models for the process. We use the [[export:/misc/publications/2012/conferences/ECML/gkronber/towerData.txt  tower data set]] which is kindly provided by Dr. Arthur Kordon from Dow Chemical (also see http://www.symbolicregression.com/?q=towerProblem). First we use symbolic regression to create a whitebox regression model for the chemical process. Afterwards we describe how the algorithm can be used to determine the most important driving factors for the process. In the end we show how symbolic regression models can be manually simplified and how visual hints can guide the user in the simplification process. 18 18 19 19 == Identification of NonLinear Models == … … 45 45 46 46 [[Image(simplifiedmodel.png, width=600)]] 47