Advertisement

Data from One Multivariate Distribution

  • Anthony C. Atkinson
  • Marco Riani
  • Andrea Cerioli
Chapter
  • 476 Downloads
Part of the Springer Series in Statistics book series (SSS)

Abstract

In this chapter we extend our analyses of the examples in Chapter 1 in order to display further features of the forward search. We use our analysis of the Swiss heads data to exemplify the properties of bivariate boxplots for data analysis. As a preparation for material on transformations of data in Chapter 4 we compare analyses of the data on national track records for women when the response is the time for the race and also its reciprocal, speed. This transformation leads to an appreciably simpler analysis. Our further analysis of the data on municipalities in Emilia-Romagna focuses on the last sixteen units to enter the forward search. For part of our analysis we reduce the data to five selected variables that explain much of the structure of the outliers. The last example is the data on Swiss bank notes. We analyse all 200 observations together and also look at the two groups separately. Forward plots of individual Mahalanobis distances, calibrated by plots of a large number of units of known origin, are shown to be a powerful tool for determining group membership.

Keywords

Mahalanobis Distance Multivariate Normal Distribution Multivariate Distribution Subset Size Estimate Covariance Matrix 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer Science+Business Media New York 2004

Authors and Affiliations

  • Anthony C. Atkinson
    • 1
  • Marco Riani
    • 2
  • Andrea Cerioli
    • 2
  1. 1.Department of StatisticsThe London School of EconomicsLondonUK
  2. 2.Dipartimento di Economia, Sezione di Statistica e InformaticaUniversità di ParmaParmaItaly

Personalised recommendations