preprocomb: Tools for Preprocessing Combinations

Preprocessing is often the most time-consuming phase in knowledge discovery and preprocessing transformations interdependent in unexpected ways. This package helps to make preprocessing faster and more effective. It provides an S4 framework for creating and testing preprocessing combinations for classification, clustering and outlier detection. The framework supports user-defined and domain-specific preprocessors and preprocessing phases. Default preprocessors can be used for low variance removal, missing value imputation, scaling, outlier removal, noise smoothing, feature selection and class imbalance correction.

Downloads: