Computational Statistics and Machine Learning: A Sparse Approach focuses on using sparse algorithms in statistics and machine learning. The first part addresses the L-0 norm minimization using greedy algorithms and considers the set covering machines, matching pursuit algorithms in machine learning, and random projection methods. The second part, which addresses L-1 norm minimization, discusses linear programming boosting, LASSO/LARS, and compressed sensing. All chapters include a detailed description of algorithms and pseudo-code and, where appropriate, a theoretical analysis of generalization ability motivating the use of sparsity. A final chapter covers applications.