Introduction to Kernel Density Estimation.
Abstract
Kernel density estimators are one technique for producing nonparametric estimates of a sample's underlying probability density function. Although there are numerous kernel functions, the reader is only introduced to the Gaussian, truncated Gaussian, mode centering Lognormal and median centering Lognormal kernels. These kernels are applied to two samples from the Fire Support Team (FIST) Force Development Testing and Experimentation II conducted by the US Army Field Artillery Board (Fort Sill, OK), at Fort Riley, KS, during April and May 1984. Analysis of the performance of the Gaussian and truncated Gaussian kernels is achieved by applying a recursive formula, developed by Richard A. Tapia and James R. Thompson, optimizing the kernel function's smoothing parameter for the FIST FDT&E II samples and for Monte Carlo simulations. Keywords: Kernel functions. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1985
- Accession Number
- ADA163920
Entities
People
- Wendy A. Winner
Organizations
- Ballistic Research Laboratory