Application of Data-Mining and Knowledge Discovery in Automotive Data Engineering

Abstract

In this paper the authors present a powerful and efficient alternative to Neural Networks (NN) by application of Knowledge Discovery and Data-Mining (KDD) methods for real world data in vehicle design, particularly for automotive Data Engineering (DE) mechanisms and processes. Typical tasks in automotive engineering are dependency analysis, classification of concepts and prediction of characteristic design parameters. From the point of view of a design engineer the main drawback of a NN-based approach is a lack of clear interpretation of the results. For classical, statistical tasks an application of an instance-based method, e.g. K-Nearest-Neighbors (KNN), represents an appropriate alternative for the engineer. By application of rule-based methods the authors demonstrate an alternate in conceptual design, which, in contrast to NN, allows to interpret the results and proof or enhance designers knowledge. The approach of this paper is based on a novel application of an Evolutionary Decision Rule Learner with Multivariate Discretization (EDRL-MD) for classification, and of M6 for regression learning.KeywordsKnowledge DiscoveryRadial Basis Function NetworkEngine RequirementEngine CapacityRegression LearningThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.