PROC RELIABILITY: DISTRIBUTION Statement

DISTRIBUTION Statement

DISTRIBUTION probability distribution-name ;

The ANALYZE, PROBPLOT, RELATIONPLOT, and MODEL statements require you to specify the probability distribution that describes your data. You can specify a probability distribution for lifetime data by using the DISTRIBUTION statement anywhere after the PROC RELIABILITY statement and before the RUN statement. If you do not specify a distribution in a DISTRIBUTION statement, the normal distribution is assumed. In addition, you can specify a parametric nonhomogeneous Poisson process model for recurrent events data in a DISTRIBUTION statement. The probability distribution for lifetime data or the model for recurrent events data specified determines the distribution for which parameters are estimated from your data. The valid distributions and the statements to which they apply are shown in Table 14.6 and Table 14.7.

Table 14.6 Probability Distributions for Lifetime Data
Distribution	Distribution-Name Specified	Statement
Binomial	BINOMIAL	ANALYZE
Exponential	EXPONENTIAL	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Extreme value	EXTREME \| EV	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Generalized gamma	GAMMA	MODEL
Logistic	LOGISTIC \| LOGIT	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Log-logistic	LLOGISTIC \| LLOGIT	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Lognormal (base $\text{[math]}$ )	LOGNORMAL \| LNORM	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Lognormal (base 10)	LOGNORMAL10 \| LNORM10	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Normal	NORMAL	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Poisson	POISSON	ANALYZE
Weibull	WEIBULL \| W2	ANALYZE, PROBPLOT,
		RELATIONPLOT, MODEL
Three-parameter Weibull	WEIBULL3 \| W3	ANALYZE, PROBPLOT

Table 14.7 Poisson Process Models for Recurrence Data
NHPP Model	Distribution-Name Specified	Statement
Homogeneous	HPP	MODEL
Crow-AMSAA	NHPP(CA) \| NHPP(CROWAMSAA)	MODEL
Log-linear	NHPP(LOG) \| NHPP(LOGLINEAR)	MODEL
Power	NHPP(POW) \| NHPP(POWER)	MODEL
Proportional intensity	NHPP(PROP) \| NHPP(PROPORTIONAL)	MODEL