H. Shore, A GENERAL FORMULA FOR THE FAILURE-RATE FUNCTION WHEN DISTRIBUTION INFORMATION IS PARTIALLY SPECIFIED, IEEE transactions on reliability, 46(1), 1997, pp. 116-121
This paper presents a new formula for the failure-rate function (FRF),
derived from a recently-introduced 4-parameter family of distribution
s. The new formula can be expressed in terms of its Cdf, is characteri
zed by algebraic simplicity, and can replace more-complex hazard funct
ions by using routine distribution fitting. When the actual Cdf is unk
nown and partial distribution-information is available (or can be extr
acted from sample data), new fitting procedures that use only first-de
gree or first- & second-degree moments are used to approximate the unk
nown FRF. This new approach is demonstrated for some commonly used Cdf
's and shown to yield highly accurate values for the FRF. Relative to
current practice, the new FRF has 4 major advantages: It does not requ
ire specification of an exact distribution, thus avoiding errors incur
red by the use of a wrong model; Since estimates of only low-degree (a
t most first- or second-degree) moments are required to determine the
parameters of the FRF, the associated mean-square-deviations are relat
ively small; The new FRF can be easily adapted for use with censored d
ata; Simple maximum likelihood estimates can be developed.