Attribute Interactions in Medical Data Analysis
- 471 Downloads
There is much empirical evidence about the success of naive Bayesian classification (NBC) in medical applications of attribute-based machine learning. NBC assumes conditional independence between attributes. In classification, such classifiers sum up the pieces of class-related evidence from individual attributes, independently of other attributes. The performance, however, deteriorates significantly when the “interactions” between attributes become critical. We propose an approach to handling attribute interactions within the framework of “voting” classifiers, such as NBC. We propose an operational test for detecting interactions in learning data and a procedure that takes the detected interactions into account while learning. This approach induces a structuring of the domain of attributes, it may lead to improved classifier’s performance and may provide useful novel information for the domain expert when interpreting the results of learning. We report on its application in data analysis and model construction for the prediction of clinical outcome in hip arthroplasty.
KeywordsDomain Expert Information Gain Negative Interaction Feature Subset Attribute Interaction
Unable to display preview. Download preview PDF.
- 1.Shapiro, A.D.: Structured induction in expert systems. Turing Institute Press in association with Addison-Wesley Publishing Company (1987)Google Scholar
- 2.Michie, D.: Problem decomposition and the learning of skills. In: Lavrač, N., Wrobel, S. (eds.) ECML 1995. LNCS, vol. 912, pp. 17–31. Springer, Heidelberg (1995)Google Scholar
- 4.Harris, W.H.: Traumatic arthritis of the hip after dislocation and acetabular fractures: Treatment by mold arthroplasty: end result study using a new method of result evaluation. J. Bone Joint. Surg. 51-A, 737–755 (1969)Google Scholar
- 5.Zupan, B., Demšar, J., Smrke, D., Božikov, K., Stankovski, V., Bratko, I., Beck, J.R.: Predicting patient’s long term clinical status after hip arthroplasty using hierarchical decision modeling and data mining. Methods of Information in Medicine 40, 25–31 (2001)Google Scholar
- 6.Jakulin, A.: Attribute interactions in machine learning. Master’s thesis, University of Ljubljana, Faculty of Computer and Information Science (2003)Google Scholar
- 12.Rish, I., Hellerstein, J., Jayram, T.: An analysis of data characteristics that affect naive Bayes performance. Technical Report RC21993, IBM (2001)Google Scholar
- 13.Demšar, J., Zupan, B.: Orange: a data mining framework. (2002), http://magix.fri.unilj.si/orange