SURVREG AFT

The SURVREG AFT command is available in the SPSS® Statistics Standard Edition or Advanced Edition.

SURVREG AFT is a command that is needed to start the parametric survival models procedure with nonrecurrent life time data. The command must be followed by either one or two numeric variables that are treated as dependent variable(s). If a single variable is specified, the survival time is represented by one variable. If two variables are specified, variable1 and variable2 will denote the lower and the upper bound of the survival time, respectively. The two variables variable1 and variable2 cannot be the same. The keyword BY is optional, which might be followed by the variable name(s) to be treated as factors. If specified, at least one variable name must be specified following it. For each factor specified, an optional value could be specified in a pair of single quotation marks within parentheses following the variable name to designate the custom last category. The user-supplied string values must be within a pair of single quotation marks but not for numeric values. If there are no cases taking on the value that is specified, the last occurring value would be used as the last category when the procedure internally encodes the factor. In the effect of TO and/or ALL used for the factor list, value that is specified in parentheses are not allowed. A variable specified following BY cannot be the same as the dependent variable(s). The keyword WITH is optional, which could be followed by the numeric variable name(s) to be treated as covariates. If specified, at least one variable name must be specified following it. A variable that is specified following WITH cannot be the same as the dependent variable(s). A variable cannot be specified by both BY and WITH. Note: At least one covariate or factor must be specified to run the procedure.

SURVREG AFT {variable}{variable1 variable2}
[ BY factor1[('value')] factor2 [('value')] ... ]
[ WITH covariate_list ]

[ /CRITERIA
[ CLASSMISSING = {EXCLUDE∗∗}{INCLUDE} ]
[ CILEVEL = {95∗∗}{value} ]
[ TREATMENT = {DISCARD∗∗}{CONFORM}{DERIVE} ]
]

[ /MODEL
[ covariate1[[value]] covariate2[[value]] ... factor1[[values]] factor2[[values]] ... ]
[ DISTRIBUTION = {WEIBULL∗∗}{EXPONENTIAL}{LOG_NORMAL}{LOG_LOGISTIC} ]
[ LEFTTRUNCATE = {variable} ]
[ SIGMA = {SE∗∗}{INV_SE}{value} ]
[ INTERCEPT = {value} ]
]

[ /STATUS
VARIABLE = {variable}
[ FAILURE = {'F'∗∗}{'value1', 'value2' ... }{NULL}
{1∗∗}{value1, value2, ... , value3 THRU value4, value5 THRU value6, ... } ]
[ RIGHT = {'R'∗∗}{'value1', 'value2' ... }{NULL}
{0∗∗}{value1, value2, ... , value3 THRU value4, value5 THRU value6, ... } ]
[ LEFT = {'L'∗∗}{'value1', 'value2' ... }
{NULL∗∗}{value1, value2, ... , value3 THRU value4, value5 THRU value6, ... } ]
[ INTERVAL = {'I'∗∗}{'value1', 'value2' ... }
{NULL∗∗}{value1, value2, ... , value3 THRU value4, value5 THRU value6, ... } ]
[ UNMAPPED = {F∗∗}{R}{L}{I}{EXCLUDE} ]
]

[ /ESTIMATION
[ SELECTFEATURES = {FALSE∗∗}{TRUE} ]
[ PENALTY = {0.001∗∗}{value} ]
[ ADMM = {FAST∗∗}{TRADITIONAL} ]
[ METHOD = {AUTO∗∗[(THRESHOLD = {1000∗∗}{integer})]}
{NEWTON_RAPHSON}
{L_BFGS}[(UPDATE = {5∗∗}{integer})]
]
[ MAXLINESEARCH = {20∗∗}{integer} ]
[ MAXITER = {20∗∗}{integer} ]
[ MAXSTEPHALVING = {5∗∗}{integer} ]
[ PCONVERGE = {0.000001∗∗}{value} [({ABSOLUTE∗∗}{RELATIVE})] ]
[ FCONVERGE = {0∗∗}{value}[({ABSOLUTE∗∗}{RELATIVE})] ]
[ HCONVERGE = {0∗∗}{value}[({ABSOLUTE∗∗}{RELATIVE})] ]
[ OPTIMIZE = {BOTH∗∗}{PRIMAL}{DUAL} ]
[ EPSILON_ABS = {0.0001∗∗}{value} ]
[ EPSILON_REL = {0.01∗∗}{value} ]
]

[ /PREDICT
[ FUTUREDURATION({variable}) ]
[ FUTUREINTERVALS(BETWEEN = {value} PERIODS = {integer}) ]
[ SURVIVAL[({PredSurvival∗∗}{var_name or root_name})] ]
[ HAZARD[({PredHazard∗∗}{var_name or root_name})] ]
[ CUMHAZARD[({PredCumHazard∗∗}{var_name or root_name})] ]
[ PASTTIME({variable}) ]
[ CONDITIONALSURVIVAL[({PredConditionalSurvival∗∗}{var_name or root_name})] ]
]

[ /PRINT
[ HISTORY[{1∗∗}{integer}] ]
[ ENCODING ]
[ INITIALS ]
]

[ /RESIDUALPLOT
[ CUTPOINTS({1000∗∗}{integer}) ]

[ /FUNCTIONPLOT
[ SURVIVAL ]
[ HAZARD ]
[ DENSITY ]
[ POINTS({100∗∗}{integer}) ]
[ PATTERN([covariate1=value] [covariate2=value] ... [factor1='value')] [factor2='value'] ...) ]
[ PLOTBY(variable) ]
[ MAX_CATEGORIES({10∗∗}{integer}) ]
]

[ /OUTFILE
{ MODEL = {'directory/filename'} }
]

** Default if the subcommand or keyword is omitted.

This command reads the active data set and causes execution of any pending commands. See the "Command Order" topic for more information.

Syntax for the SURVREG AFT extension command can be generated from the Parametric Accelerated Failure Time (AFT) Models dialog box.

Release History