Statistical learning framework is a data-driven decision making process. In general, this is an iterative process of class prediction for complex high dimensional data. Statistical learning approaches develop base prediction functions based on subsets of data (training data), final predictive function is an aggregate of all the base functions, and it is independently validated by exposing to a new data (test data). These approaches do not depend on the probabilistic distributional assumptions of data, and very adept in handing data structures characterized by small sample size, high dimensionality, and inter-dependencies. Boosting is one of the statistical learning approaches that has been used in several studies for class prediction using gene expression data. In this work, we integrate sparse sufficient dimension reduction technique with boosting, to improve the accuracy of the class prediction function.