The method of lower-bound to estimate the parameters of a Pearson type III distribution
1991
BOXIAN, WU | YU, HOU | JING, DING
A new procedure, the method of lower-bound (MLB), is proposed for determining the design quantile Xp. The basic concept is first to determine an estimate âQ of the location parameter aQ using the probability weighted moment (PWM) method, and then to transform the variable X from the original (X) space to a new (Y) space by using Y = ×—âQ. The variable Y is considered to have a two parameter gamma distribution. In Y-space, the two parameters are estimated by the PWM or an autocovariance method, then transformed from the Y-space to the X-space. Results from the Monte Carlo experiments show that the MLB estimates are less biassed than comparable moment estimates and maximum likelihood estimates, and more efficient than those of PWM for design quantiles Xp.
显示更多 [+] 显示较少 [-]