Mazzatorta Paolo, Benfenati Emilio, Lorenzini Paola, Vighi Marco
Istituto di Ricerche Farmacologiche Mario Negri Milano, Via Eritrea, 62, 20157 Milano, Italy.
J Chem Inf Comput Sci. 2004 Jan-Feb;44(1):105-12. doi: 10.1021/ci034193w.
This study deals with classification for toxicity prediction. Using a data set of 235 pesticides and 153 descriptors, we built several models using seven classification algorithms: nearest mean classifier, linear discriminant analysis, quadratic discriminant analysis, regularized discriminant analysis, soft independent modeling of class analogy, K nearest neighbors classification, classification, and regression tree. The performance of the models was then compared with the classifier, the end-points, the number of descriptor, and the diversity of the data set. Finally, we made a critical analysis of the models and descriptors.
