Nội dung được dịch bởi AI, chỉ mang tính chất tham khảo
Lựa chọn tham số điều chỉnh bằng phương pháp bootstrap
Tóm tắt
Xem xét vấn đề ước lượng θ=θ(P) dựa trên dữ liệu x_n từ một phân phối P chưa biết. Cho một họ các ước lượng T_n, β của θ(P), mục tiêu là chọn β trong I sao cho ước lượng thu được là tốt nhất có thể. Thông thường, β có thể được coi là một tham số điều chỉnh hoặc làm mịn, và việc chọn β một cách thích hợp là rất cần thiết cho hiệu suất tốt của T_n, β. Trong bài báo này, chúng tôi thảo luận lý thuyết về việc chọn β bằng phương pháp bootstrap. Cụ thể, ước lượng bootstrap của β, \hat β_n, được chọn để tối thiểu hóa một ước lượng rủi ro bootstrap thực nghiệm. Một lý thuyết tổng quát được trình bày để thiết lập tính nhất quán và các thuộc tính hội tụ yếu của các ước lượng này. Các khoảng tin cậy cho θ(P) dựa trên T_{n,\hat β_n} cũng là hợp lệ asymptotic. Nhiều ứng dụng của lý thuyết này được trình bày, bao gồm lựa chọn tỷ lệ cắt tối ưu, chọn băng thông trong ước lượng mật độ và các kết hợp tối ưu của các ước lượng.
Từ khóa
#Bootstrap #Tham số điều chỉnh #Ước lượng #Tính nhất quán #Khoảng tin cậyTài liệu tham khảo
Beran, R. (1984). Bootstrap methods in statistics,Jber. d. Dt. Math.-Verein,36, 847–856.
Beran, R. (1986). Simulated power functions,Ann. Statist.,14, 151–173.
Beran, R. and Millar, W. (1987). Stochastic estimation and testing,Ann. Statist.,15, 1131–1154.
Bickel, P. J. and Freedman, D. A. (1981). Some asymptotic theory for the bootstrap,Ann. Statist.,9, 1196–1217.
Bickel, P. and Rosenblatt, M. (1973). On some global measures of the deviations of density function estimates,Ann. Statist.,1, 1071–1095.
Billingsley, P. (1968).Convergence of Probability Measures, Wiley, New York.
Bowman, A., Hall, P. and Titterington, D. (1984). Cross-validation in nonparametric estimation of probabilities and probability densities,Biometrika,71, 341–351.
Cox, D. and Hinkley, D. (1974).Theoretical Statistics, Chapman and Hall, London.
Efron, B. (1979). Bootstrap methods: Another look at the jackknife,Ann. Statist.,7, 1–26.
Faraway, J. and Jhun, M. (1988). Bootstrap choice of bandwidth for density estimation, Tech. Report No. 157, Department of Statistics, University of Michigan.
Ghosh, M., Parr, W., Singh, K. and Babu, G. (1984). A note on bootstrapping the sample median,Ann. Statist.,12, 1130–1135.
Hall, P. and Martin, M. (1988). On bootstrap resampling and iteration,Biometrika,75, 661–671.
Hall, P., DiCiccio, T. and Romano, J. (1989). On smoothing and the bootstrap,Ann. Statist.,17, 692–704.
Jaeckel, L. (1971). Some flexible estimates of location,Ann. Math. Statist.,43, 1041–1067.
Johnstone, I. and Velleman, P. (1985). Efficient scores, variance decompositions, and Monte Carlo swindles,J. Amer. Statist. Assoc.,80, 851–862.
Léger, C. (1988). Use of the bootstrap in an adaptive statistical procedure, Tech. Report No. 296, Department of Statistics, Stanford University.
Lehmann, E. (1983).Theory of Point Estimation, Wiley, New York.
Parzen, E. (1962). On estimation of a probability density and mode,Ann. Statist.,33, 1065–1076.
Pollard, D. (1984).Convergence of Stochastic Processes, Springer, New York.
Pruitt, R. (1988). Cross-validation in the one sample location problem, Tech. Report No. 510, School of Statistics, University of Minnesota.
Romano, J. (1988a). On weak convergence and optimality of kernel density estimates of the mode,Ann. Statist.,16, 629–647.
Romano, J. (1988b). Bootstrapping the mode,Ann. Inst. Statist. Math.,40, 565–586.
Serfling, R. (1980).Approximation Theorems of Mathematical Statistics, Wiley, New York.
Sheehy, A. and Wellner, J. (1988). Uniformity inP of some limit theorems for empirical measures and processes, Tech. Report 134, Revision 2, Department of Statistics, University of Washington.
Silverman, B. (1978). Weak and strong uniform consistency of the kernel estimate of a density and its derivatives,Ann. Statist.,6, 177–184.
Stone, M. (1974). Cross-validatory choice and assessment of statistical predictions,J. Roy. Statist. Soc. Ser. B,36, 111–147.
Stone, M. (1977). Asymptotics for and against cross-validation,Biometrika,64, 29–35.