Frequentist model averaging

One of the difficulties with model selection is working out how to use a selected model and, in particular, how to use it to make valid inferences that incorporate the fact that the model has been selected.  There has been considerable recent interest in avoiding or reducing the problems of making inference after selecting a model by using one of the techniques known as model averaging.  This project will review the effects of ignoring model selection on inference, some of the frequentist methods that have been proposed for model averaging and explore some of their properties.