Data mining tasks and methods: Equation fitting: multidimensional regression analysis

  • Authors:

  • J. Sunil Rao;William J. E. Potts

  • Affiliations:

  • Assistant Professor of Biostatistics, Case Western Reserve University, Cleveland, Ohio;SAS Institute, Cary, North Carolina

  • Venue:
  • Handbook of data mining and knowledge discovery
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multidimensional regression analysis relates a target outcome Y to a vector of predictors X through a variety of possible link functions depending on the distribution of Y. The predictors may be used in a linear fashion or given a more data-driven nonparametric functional form. These variations on the modeling paradigm cover the standard linear model, generalized linear model, and generalized additive model. This article details these connections and provides algorithms for model fitting. A database-marketing example illustrates the use of multidimensional regression models.