Ventana software support forum

Hi

I have a model that uses subscripts representing different customer segmentation.
Depending on the agency, the segmentation may vary.
For instance in one agency, it is better to choose a class of customer that rent
More or less than 100 kilometer per day, in another it may be more or less than 200 kilometer a day. I would call both subscripts c1s100 and c1s200 if they represent renting one day during the week for instance.
Using the same model, if I want to switch from one agency to another I am obliged to modify the subscripts and some of the equations.
On top of that if I want to compare the results of different agencies I cannot do it because
if the model has the subscripts from one agency, the results coming from other agencies will not show. The segmentation may change too with the time.
One of the solution is to have a list of subscripts say (s1-s15) the same for all agencies, knowing that the subscript s3 for example will not have the same meaning for different agencies. It is then not necessary to modify the equations, if the sets of subscripts are correctly mapped.
But with this solution, on the results, one looses the meaning of the subscripts, one has only
s1,s2 etc… which is not practical.
Is there a way to keep at the same time the same model for different sets of subscripts, while
making it possible to see what is really behind the word s10 for instance.
The ideal would be to use aliases, the aliases being loaded from an excel spreadsheet and being different with each run.
Regards and thanks in advance.
JJ

There are no aliases for subscripts. You could use a subscript mapping, but in this case the variable names would display with the generic subscript form. You could also use literal labels for the subscript elements as in

SubscriptMeaning[s1] :IS: 'ShortTrips'

but these labels will not show up in graphs.

Another possibility would be to put all the agencies into a single model - the the subscript vector is by something like

Agency1ShortTrip,Agency1LongTrip,"Agency2cs100,Agency2 cs200

and so on. Then to switch agencies just give differnet subscripts.

For graphing you can also use custom graphs with explicits subscripts - these do not need to be valid for the current model - just the dataset being used to create the graph.

Aliases (or perhaps just labels) for subscripts - and variables in general - is an interesting notion. I will give it some thought.

Bob Eberlein

Thank's for the reply, Bob.
Regards.
JJ

We use subscript aliases in Sable, the interface package for Vensim. This makes it very quick and easy to change the name of subscript elements. The alias names appear in all Sable output (Sable's own graphics), but not in any Vensim graphs etc displayed.

The purpose of the "mapping", as it is refered to in the Sable help, is to enable rapid customisation of the model.

Many moons ago, we created a small application to read the Vensim model and replace the susbcript entries with new element names, but I can't remember exactly how it was done!

Lee

Hi Lee

Thank you for your answer.
I have the Vensim DSS package but have been working until now only with the Pro features and have still too make use of the DSS ones.
Then there will be some time until I buy Sable.
I have still to make usable models on a regularly basis which
means:
having tested and used models, having built the automatic programs that gerate the appropriate data from reality and transferring them into Excel from which they are transferred into Vensim(written in SQL) etc...
Maybe I will once use ODBC.
One thing after the other and it takes an awful long time.
So I will live with no aliases, but using Sable will certainly come one day.
Regards.
JJ