Advertisement

Error bounds in cascading regressions

  • Michael R. Karlinger
  • Brent M. Troutman
Article

Abstract

Cascading regressions is a technique for predicting a value of a dependent variable when no paired measurements exist to perform a standard regression analysis. Biases in coefficients of a cascaded-regression line as well as error variance of points about the line are functions of the correlation coefficient between dependent and independent variables. Although this correlation cannot be computed because of the lack of paired data, bounds can be placed on errors through the required properties of the correlation coefficient. The potential meansquared error of a cascaded-regression prediction can be large, as illustrated through an example using geomorphologic data.

Key words

cascaded regressions error bounds population parameters channel morphology 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Graybill, F. A., 1976, Theory and application of the linear model, Duxbury Press, N. Scituate, Mass., 709 p.Google Scholar
  2. Jones, D. A., 1983, Statistical analysis of empirical models fitted by optimization: Biometrika, v. 70, no. 11, p. 67–88.Google Scholar
  3. Schumm, S. A., 1968, River adjustment to altered hydrologic regimen—Murrumbidgee River and paleochannels, Australia: U.S. Geological Survey Professional Paper 598, 65 p.Google Scholar
  4. White, H., 1981, Consequences and detection of misspecified nonlinear regression models: Jour. Amer. Stat. Assoc., v. 76 (374), p. 419–433.Google Scholar
  5. Williams, G. P., 1983, Improper use of regression equations in the earth sciences: Geology, v. 11, p. 195–197.Google Scholar

Copyright information

© Plenum Publishing Corporation 1985

Authors and Affiliations

  • Michael R. Karlinger
    • 1
  • Brent M. Troutman
    • 1
  1. 1.U.S. Geological SurveyDenver Federal CenterDenverUSA

Personalised recommendations