This paper describes our work on the Sisyphus challenge dataset, which includes both classification and clustering tasks. We present our work in the context of the CRISP-DM methodology. Further key aspects of the work are the evaluation and integration of multiple models by means of ROC analysis. We indicate a simple method of forcing classifiers to cover the whole of the ROC space. In conclusion, we outline several promising research directions.
|Translated title of the contribution||Data Mining on the Sisyphus Dataset: Evaluation and Integration of Results|
|Title of host publication||Unknown|
|Editors||Christophe Giraud-Carrier, Nada Lavrac, Steve Moyle|
|Publisher||ECML/PKDD'01 workshop notes|
|Pages||69 - 80|
|Number of pages||11|
|Publication status||Published - Sep 2001|