TY - GEN

T1 - Computer implementation of probability distribution quantile estimation

AU - Yu, Xian Chuan

AU - Yuan, Zhongyi

AU - Yu, Chen

AU - Yang, Meng

PY - 2005/12/12

Y1 - 2005/12/12

N2 - Generally, we get probability distribution quantile by looking through numerical tables, however, it is not only easy to make mistake, but also limited in precision, no more than 0.0001. And programming techniques up to now are either too restrictive to be applied to general cases, or too complicated to be implemented for practical use. Therefore, there is a need for robust procedures to estimate quantiles, which can be applied to relatively generic processes and easy to implement. The paper briefly discusses the algorithm and the implement of some familiar probability distribution quantiles, such as, standardized normal distribution, β distribution, X2 distribution, t distribution and F distribution. Especially, we use Newton dichotomy here to improve the precision, in the case of t and F distributions which is insufficient by approximate formulae only, because of the accumulated error. An experimental performance evaluation demonstrates the validity of these procedures to calculate probability distribution quantiles.

AB - Generally, we get probability distribution quantile by looking through numerical tables, however, it is not only easy to make mistake, but also limited in precision, no more than 0.0001. And programming techniques up to now are either too restrictive to be applied to general cases, or too complicated to be implemented for practical use. Therefore, there is a need for robust procedures to estimate quantiles, which can be applied to relatively generic processes and easy to implement. The paper briefly discusses the algorithm and the implement of some familiar probability distribution quantiles, such as, standardized normal distribution, β distribution, X2 distribution, t distribution and F distribution. Especially, we use Newton dichotomy here to improve the precision, in the case of t and F distributions which is insufficient by approximate formulae only, because of the accumulated error. An experimental performance evaluation demonstrates the validity of these procedures to calculate probability distribution quantiles.

UR - http://www.scopus.com/inward/record.url?scp=28444495970&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=28444495970&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:28444495970

SN - 078039092X

SN - 9780780390928

T3 - 2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005

SP - 2783

EP - 2788

BT - 2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005

T2 - International Conference on Machine Learning and Cybernetics, ICMLC 2005

Y2 - 18 August 2005 through 21 August 2005

ER -