Fence Methods

This book is about a recently developed class of strategies, known as the fence methods, which fits particularly well in non-conventional and complex model selection problems with practical considerations.

The idea involves a procedure to isolate a subgroup of what are known as correct models, of which the optimal model is a member.

This is accomplished by constructing a statistical fence, or barrier, to carefully eliminate incorrect models.

Once the fence is constructed, the optimal model is selected from amongst those within the fence according to a criterion which can be made flexible.

In particular, the criterion of optimality can incorporate consideration of practical interest, thus making model selection a real life practice.Furthermore, this book introduces a data-driven approach, called adaptive fence, which can be used in a wide range of problems involving determination of tuning parameters, or constants.

Instead of relying on asymptotic theory, the fence focuses on finite-sample performance, and computation.

Such features are particularly suitable to statistics in the new era.Contents:IntroductionExamplesAdaptive FenceRestricted FenceInvisible FenceFence Methods for Small Area Estimation and Related TopicsShrinkage Selection MethodsModel Selection with Incomplete DataTheoretical PropertiesReadership: Graduates and researchers interested in a new class of strategies for model selection.Key Features:Introduces a general data-driven procedure, and let the data speak in choosing some critical tuning constantsTargets non-conventional and complex problems, and focuses on finite-sample performance and computationMakes model selection a real life practice