Journal of the Korean Statistical Society

The Korean Statistical Society (한국통계학회)
권호 표지

Model Selection for Tree-Structured Regression

  • Published : 1996.03.01
Download PDF
⟨ Previous Next ⟩

Abstract

In selecting a final tree, Breiman, Friedman, Olshen, and Stone(1984) compare the prediction risks of a pair of tree, where one contains the other, using the standard error of the prediction risk of the larger one. This paper proposes an approach to selection of a final tree by using the standard error of the difference of the prediction risks between a pair of trees rather than the standard error of the larger one. This approach is compared with CART's for simulated data from a simple regression model. Asymptotic results of the approaches are also derived and compared to each other. Both the asymptotic and the simulation results indicate that final trees by CART tend to be smaller than desired.

Keywords

References

  1. Discrete Multivariate Analysis: Theory and Practice Bishop, Y. M. M.;Fienberg, S. E.;Holland, P. W.
  2. Classification and Regression Trees Breiman, L.;Friedman, J. H.;Olshen, R. A.;Stone, C. J.
  3. Operational Res. Quart. v.24 The use of automatic interaction detector and similar search procedures Doyle, R. M.
  4. Pub. Op. Quart. v.36 Alchemy in the behavioral sciences Einhorn, H.
  5. Communications in Statistics Theory and Methods v.23 no.4 A general property among nested, pruned subtrees of a decision-support tree Kim, S. H.
  6. Journal of American Statistical Association v.83 Tree-structured classification via generalized discriminant analysis Loh, W. Y.;Vanichesetakul, N.
  7. THAID: a sequential search program for the analysis of nominal scale dependent variables Morgan, J. N.;Messenger, R. C.
  8. Journal of American Statistical Association v.58 Problems in the analysis of survey data, and a proposal Morgan, J. N.;Sonquist, J. A.
  9. Linear Statistical Inference and Its Applications(2nd ed.) Rao, C. R.