Access

You are not currently logged in.

Access your personal account or get JSTOR access through your library or other institution:

login

Log in to your personal account or through your institution.

If you need an accessible version of this item please contact JSTOR User Support

Adaptive Regression by Mixing

Yuhong Yang
Journal of the American Statistical Association
Vol. 96, No. 454 (Jun., 2001), pp. 574-588
Stable URL: http://www.jstor.org/stable/2670298
Page Count: 15
  • Download ($14.00)
  • Cite this Item
If you need an accessible version of this item please contact JSTOR User Support
Adaptive Regression by Mixing
Preview not available

Abstract

Adaptation over different procedures is of practical importance. Different procedures perform well under different conditions. In many practical situations, it is rather hard to assess which conditions are (approximately) satisfied so as to identify the best procedure for the data at hand. Thus automatic adaptation over various scenarios is desirable. A practically feasible method, named adaptive regression by mixing (ARM), is proposed to convexly combine general candidate regression procedures. Under mild conditions, the resulting estimator is theoretically shown to perform optimally in rates of convergence without knowing which of the original procedures work the best. Simulations are conducted in several settings, including comparing a parametric model with nonparametric alternatives, comparing a neural network with a projection pursuit in multidimensional regression, and combining bandwidths in kernel regression. The results clearly support the theoretical property of ARM. The ARM algorithm assigns weights on the candidate models-procedures via proper assessment of performance of the estimators. The data are split into two parts, one for estimation and the other for measuring behavior in prediction. Although there are many plausible ways to assign the weights, ARM has a connection with information theory, which ensures the desired adaptation capability. Indeed, under mild conditions, we show that the squared L2, risk of the estimator based on ARM is basically bounded above by the risk of each candidate procedure plus a small penalty term of order l/n. Minimizing over the procedures gives the automatically optimal rate of convergence for ARM. Model selection often induces unnecessarily large variability in estimation. Alternatively, a proper weighting of the candidate models can be more stable, resulting in a smaller risk. Simulations suggest that ARM works better than model selection using Akaike or Bayesian information criteria when the error variance is not very small.

Page Thumbnails

  • Thumbnail: Page 
574
    574
  • Thumbnail: Page 
575
    575
  • Thumbnail: Page 
576
    576
  • Thumbnail: Page 
577
    577
  • Thumbnail: Page 
578
    578
  • Thumbnail: Page 
579
    579
  • Thumbnail: Page 
580
    580
  • Thumbnail: Page 
581
    581
  • Thumbnail: Page 
582
    582
  • Thumbnail: Page 
583
    583
  • Thumbnail: Page 
584
    584
  • Thumbnail: Page 
585
    585
  • Thumbnail: Page 
586
    586
  • Thumbnail: Page 
587
    587
  • Thumbnail: Page 
588
    588