Activity Number:
|
71
|
Type:
|
Contributed
|
Date/Time:
|
Monday, August 12, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Nonparametric Statistics*
|
Abstract - #300808 |
Title:
|
Non-parametric Density Estimation Via Wavelets
|
Author(s):
|
Ren Zhang*+ and Lawrence Brown
|
Affiliation(s):
|
University of Pennsylvania and University of Pennsylvania
|
Address:
|
3000 SH-DH, Univ. of Pennsylvania, Philadelphia, Pennsylvania, 19104, US
|
Keywords:
|
Non-parametric ; wavelet ; density ; root-unroot
|
Abstract:
|
A new methodology for the application of wavelets in non-parametric density estimation is proposed. It transfers a density estimation problem into a regression problem by binning the observations and then treating the square root of the observation counts as the new data for regression. It then uses a wavelet regression method to recover the square root of the density. Because of the automatic adaptivity of wavelets methods, this density estimation method achieves the optimal convergence rate and is computationally efficient. Data from the call service center of a large northeast bank is used to demonstrate the usage of this method for practical problems. In this setting, the density estimator is used to describe the arrival rate of the phone call as a function of covariates such as time-of-day and day of the week.
|