Survival Kit, software/Older versions

Before you begin

edit

First you should download the source codes, examples and manual from boku.ac.at. Everything is for free, programs are open source. The manual in currently in form of a .ps file. If you don't have a program able to open .ps files you have to download one (e.g. Postscript Viewer) or convert it to .pdf (with e.g. ps2pdf). You are encouraged to Google for more possibilities.

Compilation

edit

The word "compilation" means turning the source code (the Fortran program in text form) into an executable file which may be run by the operating system (in Windows .exe).
The Fortran source codes need to be compiled before usage. The possible options for this include (but not limited to) Lahey, gfortran, Force compilers. You should also recompile the programs after any change in the source code.
The necessity of (re)compilation is one of the major difficulties in using the Survival Kit v3.12 especially for novice users. This was required because the programs were initially written in Fortran77 allowing only matrices of fixed size. In the upcoming version the memory allocation possibilities of Fortran90 are used which makes the programs much more flexible.

Suggestion

edit

In order to avoid at least some of the necessary recompilation, you may set/change the parameters in the "parinclu" file before the first compilation as follows:

   PARAMETER (SUPPLY_INPUT0=.true.)
   PARAMETER (SUPPLY_INPUT1=.true.)
   PARAMETER (SUPPLY_INPUT2=.true.)
   PARAMETER (SUPPLY_OUTPUT=.true.)

Simply open the parinclu file using your favorite text editor, modify the parameters and save. The outcome of this (after the compilation) will be that you will need to type the file names at running time, but it also gives more flexibility when the input file names are changed. In this description we will assume that you have done this, i.e. set all parameters to .true.


Small example

edit

Directory

edit

First thing you need to do is to copy small.dat, small.par, small.cox and small.wei along with the compiled executable files prepare.exe, weibull.exe and cox.exe. One of the common mistakes which novice users make is to keep multiple evaluations in one directory. This is not a technical problem, but after several runs the directory is filled with the results and additional files created by the Survival Kit which may lead to confusions and a big mess. Keep separate databases in separate directories. This is a practical experience...

Run Prepare

edit

To do this you should start the program in the command prompt. Click Start then Run.. and write "cmd" and hit Enter. In the command prompt navigate to the directory with the prepare.exe file and run it. You will be asked for the name of the parameter file, write: small.par

If you get an error message saying something like "The value of the STATUS specifier in an OPEN statement..." then you have typed the file name for the parameter file incorrectly. If the program does not ask you about the file name, then the parameter SUPPLY_INPUT0 is set to .false. in parinclu. In this case you may change the name of your file to the one indicated in INPUT_FILE0 or change its value to .true. in parinclu and recompile the program (as explained here).

Run Weibull

edit

You can run the Weibull program just after running the prepare.exe at least once, since it needs some files that are created by the Prepare program. These additional files contain the recoded values for your variables. You can find out more information about them in the manual, but please note that you don't have to modify them at all.

You should run the Prepare program before running Weibull.It doesn't mean however that you have to run Prepare every time. If you change the MODEL or adjust different settings you do not need to rerun Prepare again if every variable needed in the Weibull program is included in Prepare's OUTPUT statement. Please note that not every variable from this OUTPUT statement should be used in the MODEL (i.e. you can include all effects to the OUTPUT in Prepare and use just a few in Weibull's MODEL).
If you run stratified analysis in Weibull the output data file from prepare (small2.dat in this example) should be sorted ascending by the STRATA variable.

When running the Weibull program you will be asked for input file names. Again "The value of the STATUS specifier in an OPEN statement..." error message may pop up here from the same reason as for Prepare. If this happens, check if the file names are entered correctly. The default entries for the "Small example" are:

first parameter file (=output from prepare) small.paa
second parameter file (=model specification) small.wei
detailed output file any_name_you_want or just_hit_enter

Notice the file extensions of .paa (contains the list of variables) and .wei (contains your specifications of the model you want to use). These are the default filename settings which may be altered changing parameters SUPPLY_INPUT1 and SUPPLY_INPUT2 in parinclu. Remember: after each change in the source code you have to recompile the programs.

Run Cox

edit

Running the Cox program is very similar to Weibull. The major difference is that you have to sort the recoded data file (small2.dat in this example) ascending by time variable (1st column) and descending by censoring variable (2nd column). If you run a stratified analysis the sorting order modifies to: ascending by strata variable (the column number of the strata), ascending by time variable (1st column) and descending by censoring variable (2nd column).

The default entries when running the "Small example" are:

first parameter file (=output from prepare) small.paa
second parameter file (=model specification) small.cox
detailed output file any_name_you_want or just_hit_enter

Where the file with extension .cox contains the specifications of your Cox model.