# NYU: Sparsity of Mallows Model Averaging Estimator

A study co-authored by Dr. Yang Feng, associate professor in biostatistics at the New York University School of Global Public Health, was published by Economics Letters titled “On the sparsity of Mallows model averaging estimator.”

Findings show that Mallows model averaging estimator proposed by Hansen (2007) can be written as a least squares estimation with a weighted L1 penalty and additional constraints. This Mallows Model Averaging has an equivalent constrained weighted Lasso formulation. By exploiting this representation, the weight vector obtained by this model averaging procedure has a sparsity property in the sense that a subset of models receives exactly zero weights. This means that the solution of Mallows Model Averaging is sparse in finite settings. Moreover, this representation allows researchers to adapt algorithms developed to efficiently solve minimization problems with many parameters and weighted L1 penalty. In particular, the authors developed a new coordinate-wise descent algorithm for model averaging. Simulation studies show that the new algorithm computes the model averaging estimator much faster and requires less memory than conventional methods when there are many models.

Tags: ,