Approximate Distributions of Weighted Sum of Correlated Chi-square Random Variables
博士 === 國立中正大學 === 應用數學研究所 === 97 === Many authors have discussed the approximate distribution of a weighted sum of mutually independent chi-squared random variables. The results of previous studies can be used to combining independent tests of significance. In some situations, the individual tests a...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Online Access: | http://ndltd.ncl.edu.tw/handle/h44e37 |
| Summary: | 博士 === 國立中正大學 === 應用數學研究所 === 97 === Many authors have discussed the approximate distribution of a weighted sum
of mutually independent chi-squared random variables. The results of previous
studies can be used to combining independent tests of significance. In some situations,
the individual tests are correlated. Thus the test statistic is no longer
a weighted sum of independent chi-squared variables. This paper studies the
approximate distribution of a weighted sum of correlated or uncorrelated chisquared
variables. This approximate distribution can be obtained by two types
of approximations, namely the chi-squared approximation and the normal approximation.
This study calculates the error bounds of both approximations, and
compares using simulations.
|
|---|