Title:   Model Building and Diagnostics for Massive Data Sets


Abstract:

The process of statistical modeling is an iterative process,
which culminates with model modification if indicated by
failure in the diagnostics phase.  This process of model
modification is an art, but perhaps still poorly understood
and difficult to teach.  We discuss the extra difficulties
in model diagnostics and modification resulting from the
extraordinary burdens of working in the world of massive data sets.