<< >> Contents

8.11 Using the Output Processor

A vital part of any simulation study is analysing the results of the experimental work on the model and preparing the results for publication in a report. In ARENA the output processor is responsible for these functions.

From the main menu select `Output' and from the submenu displayed select either `New' or `Load'. `Load' will load a previous output session. Selecting `New' will open a new `Data Group' window. The first task is to import into the data group files for analysis. These files will have been generated using options available in the `DSTATS' or `TALLIES' blocks (see 8.10). In the new window select `Files' and then `Add' to bring the required files into the data group. Although any file extension can be used for your data files the advantage of using ".DAT" is now apparent because ARENA automatically lists files with this extension. Load in the appropriate file or files. You are now ready to display and analyse this data. The `Display' menu allows you a choice of displaying the data as `Barchart', `Histogram', `MovAverage', `Plots' or `Tables'. Perhaps a good place to start is to to display time persistent data using `Plots'. When this item is selected a set of display options is made available (e.g. axis labels, display ranges etc.). Our advice in the first instance is simply to select the default values by immediately choosing `Accept' Having viewed the data in this way and confirmed that valid data has been obtained feel free to experiment with other options, display ranges etc. The third menu item `Statistics' contains functions too advanced for this elementary introduction but if you are familiar with basic statistical theory you will find the analysis features available very powerful.

8.11.1 Obtaining Output

You will, of course want to print out your output for reports and display purposes. A word of warning here, the print facilities available in this version of ARENA are very poor indeed. All you will obtain is a `pixel dump' of the screen display. The quality is poor and the printing time very long. The best technique for obtaining output is to export the data to a text file which can then be read by a good display package (e.g. Microsoft EXCEL). To do this select from the `Files' menu `Filter'. This will filter the data and prepare a copy of the data file for export with the file extension ".FLT". Once selected a list of filter options is presented. The most important is the `Type of batching' which can be either `TIMe' or `OBServation' (enter the first three letters). If you select `OBS' and then select a group of `5' your observations will be grouped in sets of 5 and the average of each set sent to the output file. In other words the filter function smoothes out the data and (in this example) reduces the number of observations by a factor of 5. Batching by time achieves a similar result but the number entered in `Obs or time per group' is interpreted as the time interval over which to average the data. You may also wish to truncate the data by removing a set of initial observations which are part of the `warm up' or non equilibrium region of the experiment. Again you can truncate by OBS or by TIM. After selecting accept the filter process will take place and you will be given a brief summary report of the result.

Now go back to the `Files' menu and select `Export>>' and `Export' from the submenu displayed. This will generate a text file of the exported data with the file extension ".DAX". This file you can examine, and modify, with any text editor and import into any display package.

When you have finished using the output processor you can close it in the normal manner (`clicking' on the `A' symbol). You are offered the choice of saving the data group you have created by supplying a data group name. This is useful if you wish to return to the same data group later (perhaps after changing some experiment parameters) for further analysis.


<< >> Contents