EEBoost: A general method for prediction and variable selection based on estimating equations

The modern statistical literature is replete with methods for performing variable selection and prediction in standard regression problems. However, simple models may misspecify or fail to capture important aspects of the data generating process such as missingness, correlation, and over/underdispersion. This realization has motivated the development of a large class of estimating equations which account for these data characteristics and often yield improved inference for lowdimensional parameters. In this paper we introduce EEBoost, a novel strategy for variable selection and prediction which can be applied in any problem where inference would typically be based on estimating