Discovery of toxicological patterns with lazy learning

In this paper we propose the use of a lazy learning technique called LID for discovering patterns in the Toxicology dataset. LID classifies examples and builds an explanation of that classification. We analyzed the Toxicology dataset using a two-step proces: first we use LID for classifying all the cases in the dataset. Then we select a subset of explanations and use them as patterns that capture structural regularities (patterns) among carcinogenic chemical compounds.