Nonparametric Probability Density Estimation for Data Analysis in Several Dimensions
Abstract
The purpose of this paper is to illustrate how nonparametric probability density estimates, in particular the corresponding contour curves, are a useful adjunct to scatter diagrams when performing a preliminary examination of a set of random data in several dimensions. For a preliminary approach we generally want to perform fairly simple tasks with free-form techniques to uncover structures and features of interest in the data. Such procedures are often graphical and unlike summary statistics seldom lead to much compression of the data. Tukey presents a wealth of such procedures. One which well illustrates the power and flexibility of these preliminary procedures is the running median smoothing algorithm for time series data (with resmoothing of the rough and the like). Other graphical techniques for multivariate data are presented in Tukey and Tukey.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1983
- Accession Number
- ADP001586
Entities
People
- David W Scott
Organizations
- Rice University