The Effect of Outliers on the Regression Analysis Results with Application on the Premature Births

A. Younis, Farah; A. Dabdoub, Marwan

doi:10.33899/rjs.2006.43333

	The Effect of Outliers on the Regression Analysis Results with Application on the Premature Births
Rafidain Journal of Science
Article 7, Volume 17, Issue 2, 2006, Pages 62-81 PDF (0 K)
DOI: 10.33899/rjs.2006.43333
Authors
Farah A. Younis; Marwan A. Dabdoub
Abstract
The outliers were founded as results of the regression analysis and also by box plot graph. The outlier values were tackled by trimmed mean. The data about the life of premature births and other nine factors were collected from three records which represented by pre-blockade years 89-1990, the years 94-1995 which are under blockade and before memorandum of understanding and after memorandum of understanding the years 99-2000. A random sample of size 100 premature births were taken from each period. It was found that the outliers affected regression analysis results, so it is necessary to search for outliers whenever regression analysis is needed. The pregnancy period and the weight of premature had the major effect on his life in pre-blockade and after memorandum of understanding periods, before the period of memorandum of understanding two factors were added they were: period of nursery stay and type of disease that the premature births suffered from. The abortion cases increase in last period.

Statistics Article View: 333 PDF Download: 534