Extreme values from skewed distributions

被引：6

作者：

Ratz, HC ^{[1
]}

机构：

[1] Univ Waterloo, Dept Elect & Comp Engn, Waterloo, ON N2L 3G1, Canada

来源：

IEEE TRANSACTIONS ON EDUCATION | 2000年 / 43卷 / 04期

关键词：

calculated risk; extreme values; rare event probabilities; skewed data outliers;

D O I：

10.1109/13.883349

中图分类号：

G40 [教育学];

学科分类号：

040101 ; 120403 ;

摘要：

Data from observations that cannot have negative values, such as pollution concentrations, traffic counts, or time intervals, are generally asymmetric with positive skew, If problems are assigned involving confidence limits, or the chances of rare events, with such skewed data, then the use of Gaussian-based statistics can result in substantial error. The magnitude of the error depends upon the mean value of the data expressed in units of the standard deviation. A set of curves is given which provides a method for correcting the Gaussian based tests, or at least for estimating the possible magnitude of the error.

引用

页码：400 / 402

页数：3