how to set a target value for my model?

[SP-Felix]

Hello and good day everyone!

First off: I used the search function but I couldn't find the answer i was looking for.

So if this has been answered already i apologize!

So here is what I am trying to do:

I am building a model at the moment, with a couple of variables that in the end point to the 3 variables "churn", "number of new customers" and "average annual recurring revenue"

now these 3 all point to the final variable "growth of revenue".

I can run simulations and everything, so I guess so far my model is ok. However, if I dont want to play around with variables anymore and just set a targeted value for the "growth of revenue" variable, how would I do that?
Ideally, I'd also like to be able to understand which of the other variables have the biggest impact on the "growth of revenue".


I have attached a screenshot of my model, in case you couldnt imagine it from my writing. I hope this is sufficient info for you, if not, or if you want me to upload the model instead, please let me know.

I am really looking forward to some input from you guys and every bit of information is highly appreciated!

[tomfid]

Are you looking for a dynamic target that affects the behavior of decisions, or goal-seeking optimization?

[SP-Felix]

Hi tomfid, thanks for your reply.

I think goal-seeking optimization is what I'm looking for. You got an idea on how I can do this?

[Administrator]

I can run simulations and everything, so I guess so far my model is ok.

Hopefully. Just because it runs does not mean much though, it's good practice to make sure the model is dimensionally correct, a mass-balance check also gives others confidence that the model does not contain any of the common errors that creep in.

However, if I dont want to play around with variables anymore and just set a targeted value for the "growth of revenue" variable, how would I do that?

Do you mean let Vensim adjust inputs so that the "growth of revenue" meets a pre-defined limit? If yes, then that's optimisation (not available in PLE). You can play around with the model in SyntheSim mode though to try and meet the target.

Ideally, I'd also like to be able to understand which of the other variables have the biggest impact on the "growth of revenue".

For this you would need to use sensitivity/Monte-Carlo simulations (not available in PLE). This runs the model many times varying some of the inputs to see the effects on outputs.

[SP-Felix]

I can run simulations and everything, so I guess so far my model is ok.

Hopefully. Just because it runs does not mean much though, it's good practice to make sure the model is dimensionally correct, a mass-balance check also gives others confidence that the model does not contain any of the common errors that creep in.

Thanks for the info, ill make sure to do just that.

However, if I dont want to play around with variables anymore and just set a targeted value for the "growth of revenue" variable, how would I do that?

Do you mean let Vensim adjust inputs so that the "growth of revenue" meets a pre-defined limit? If yes, then that's optimisation (not available in PLE). You can play around with the model in SyntheSim mode though to try and meet the target.

Yeah, that's basically what I had in mind. Would the other procedure you mentioned, with the dynamic target that affects the behaviour of decisions be possible with PLE? Just asking out of interest

Ideally, I'd also like to be able to understand which of the other variables have the biggest impact on the "growth of revenue".

For this you would need to use sensitivity/Monte-Carlo simulations (not available in PLE). This runs the model many times varying some of the inputs to see the effects on outputs.

Ok great, thanks for letting me know!

[tomfid]

Assuming the latter, you could do the following:
- Create a metric that is a function of the gap between your target and actual values, like
error = (target-actual)^2
weight = {some negative value}
- Create a payoff that minimizes the metric
*p
error/weight
- Optimize

For this to work, the model and control file need to specify the constraints on decisions robustly.