Data to be analysed can be in 1 or 2 file(s). The data file must containing information on the trait(s) data
to be analyzed together with class variables and covariables for each model must be provided.
The data file(s) can be in ASCII or BINARY format with one record per observation. Any program can be used
to create this file. If ASCII format is used, all integers and reals in the data file are checked for legal
range for INTEGER*4 and REAL*4 resp.
If BINARY format is used it must follow the rules for unformatted FORTRAN input on the particular
installation. In this case it is the user's responsibility to insure, that data values do not exceed the
range for INTREGER*4 and REAL*4 respectively.
If one data file is used, each record in the data file must first consist of integer input in INTEGER*4
format followed by real input in REAL*4 format. If two data files are used the integer variables of a record
must be in the first file, and the corresponding real variables must be in the second file. The two data
files must be in the same format type.
The integer input consists of all class variables in the model(s). The real input consists of the values
for trait(s) to be analyzed and all necessary covariables and weight variables. In integer input a zero (0)
is treated as missing whereas real input below the value ”miss” specified on the $DATA keyword line in the
driver file is recognized as a missing value.
The data files can contain variables that are not used in a particular run of DMU.
Except for the requirement of integer variables before real variables or that, the two types are in
different files, there is no restriction on the order of the variables in the data files. The variables
on the input data set(s) are regarded as a list of integer input and a list of real input. If variable
names are specified in the driver file, names and variables are associated so first integer variable gets
the first name specified; integer variable number 2 gets the second variable name and so on. If variable
names are supplied, names for all variables in the input data must be specified. If no variable name
is specified, the variable is named I1-I(number of integer input) and R1-R(number of real input)