EXACT DISTRIBUTION OF SQUARED WELSCH-KUH DISTANCE AND IDENTIFICATION OF INFLUENTIAL OBSERVATIONS
Keywords:
Squared Welsch-Kuh Distance Measures, Influential Observation, Series Expression Form, Gauss Hyper-Geometric Function, Mean, Variance, Critical Points.Abstract
This paper proposes the exact distribution of squared DFFITS alias squared Welsch- Kuh ( 2 ) WK distance measure used to evaluate the influential observations in a multiple linear regression analysis. The authors have explored the relationship between the 2 WK in terms of two independent F-ratio’s and they have shown the derived density function of the 2 WK distance in a complicated series expression form involving Gauss hyper-geometric function with two shape parameters p and n. Moreover, the mean, variance of the distribution are derived in terms of the shape parameters and the authors have established the upper control limit of 2 WK . Similarly, the critical points of squared Welsch-Kuh ( 2 ) WK distance measure are computed at 5% and 1% significance levesl for different sample sizes and varying no. of predictors. Finally, the numerical example shows the identification of the influential observations and the results extracted from the proposed approaches are more scientific, systematic and their exactness outperforms the Welsch-Kuh’s traditional approach.
Downloads
References
Ali, S. Hadi (1992). A new measure of overall potential influence in linear
regression, Computational Statistics & Data Analysis, 14(1), p. 1-27.
Behnken, D. W. and Draper, N. R. (1972). Residuals and Their Variance
Patterns, Technometrics, 14, p. 101-111.
Belsley, D.A., Kuh, E. and Welsch, R.E. (1980). Regression Diagnostics:
Identifying Influential Data and Sources of Collinearity, Wiley, New York.
Bollen, K. A. and Jackman, R. W. (1990). Regression diagnostics: An
expository treatment of outliers and influential cases, Chapter 6, Modern
Methods of Data Analysis, p. 257-291.
Cook, R.D. (1977). Detection of influential observation in linear regression.
Technometrics, 19, p. 15-18.
Chatterjee, S. and Hadi, A. S. (1988). Sensitivity Analysis in Linear
Regression, New York: John Wiley and Sons.
Cook, R. D. and Weisberg, S. (1982). Residuals and influence in
regression (Vol. 5). New York: Chapman and Hall.
Dı́az-Garcı́a, J. A., and González-Farı́as, G. (2004). A note on the Cook's
distance, Journal of Statistical Planning and Inference, 120(1), p. 119-136.
Eubank, R.L (1985). Diagnostics for smoothing splines, J. Roy. Statist. Soc.,
Ser. B, 47, p. 332–341.
Hoaglin, D.C. and Welsch, R.E. (1978). The Hat matrix in regression and
ANOVA, The Amer. Statist., 32, p. 17-22.
Kim, C. (1996). Cook’s distance in spline smoothing, Statist. Probab. Lett.,
, p.139–144.
Kim, C. and Kim, W. (1998). Some diagnostics results in nonparametric
density estimation, Comm. Statist. Theory Methods, 27, p. 291–303.
Kim, C., Lee, Y. and Park, B.U. (2001). Cook’s distance in local polynomial
regression, Statist. Probab. Lett., 54, p. 33–40.
Silverman, B.W. (1985). Some aspects of the spline smoothing approach to
non-parametric regression curve fitting (with discussion), J. Roy. Statist.
Soc., Ser. B, 47, p. 1–52.
Thomas, W. (1991). Influence diagnostics for the cross-validated smoothing
parameter in spline smoothing, J. Amer. Statist. Assoc., 86, p. 693–698.
Weisberg, S. (1980). Applied linear regression. New York: Wiley.
Welsch, R. E., and Kuh, E. (1977). Technical Report 923-77: Linear
Regression Diagnostics, Cambridge, MA: Sloan School of Management,
Massachusetts Institute of Technology.
Welsch, R. E. and Peters, S.C. (1978). Finding influential subsets of data in
regression models, Proc. Eleventh Inference Symp, Comput.Sci. Statist, p.
-244.
