We propose a procedure for computing a fast approximation to regression estimates based on the minimization of a robust scale. The procedure can be applied with a large number of independent variables where the usual algorithms require an unfeasible or extremely costly computer time. Also, it can be incorporated in any high-breakdown estimation method and may improve it with just little additional computer time. The procedure minimizes the robust scale over a set of tentative parameter vectors estimated by least squares after eliminating a set of possible outliers, which are obtained as follows. We represent each observation by the vector of changes of the least squares forecasts of the observation when each of the data points is deleted. Then we obtain the sets of possible outliers as the extreme points in the principal components of these vectors, or as the set of points with large residuals. The good performance of the procedure allows identification of multiple outliers, avoiding masking effects. We investigate the procedure's efficiency for robust estimation and power as an outlier detection tool in a large real dataset and in a simulation study.
The Journal of the American Statistical Association (JASA) has long been considered the premier journal of statistical science. Science Citation Index reported JASA was the most highly cited journal in the mathematical sciences in 1991-2001, with 16,457 citations, more than 50% more than the next most highly cited journals. Articles in JASA focus on statistical applications, theory, and methods in economic, social, physical, engineering, and health sciences and on new methods of statistical education.
Building on two centuries' experience, Taylor & Francis has grown rapidlyover the last two decades to become a leading international academic publisher.The Group publishes over 800 journals and over 1,800 new books each year, coveringa wide variety of subject areas and incorporating the journal imprints of Routledge,Carfax, Spon Press, Psychology Press, Martin Dunitz, and Taylor & Francis.Taylor & Francis is fully committed to the publication and dissemination of scholarly information of the highest quality, and today this remains the primary goal.
This item is part of JSTOR collection
For terms and use, please refer to our Terms and Conditions
Journal of the American Statistical Association
© 1999 American Statistical Association
Request Permissions
