Dependent Variable Noise

Use this forum to discuss any issues relating to Systems Dynamics and Systems Thinking.
Post Reply
johnjohnson4
Member
Posts: 47
Joined: Mon Mar 14, 2016 4:58 pm
Vensim version: PLE+

Dependent Variable Noise

Post by johnjohnson4 »

I am doing a design of experiment with my model and need to add "pure error" ( i.e., noise) to a dependent variable. The variable is IRE Trend (∆I/∆Itp) in the attached model. Is there a prefered or recomended way to add noise to a variable?
a) Multiply by the dependent variabe by a random variabl
b) Add a randome variable to the dependent variable
c) Other methods

Is there partticular value in using the "Pink noise" function vs another randome or uniform disribution?

Thanks in advance,
john
Attachments
EngineeredSystem_Simulation_08072016_FINAL_J2.mdl
(11.24 KiB) Downloaded 275 times
bffcosta
Junior Member
Posts: 3
Joined: Mon Aug 08, 2016 10:40 pm
Vensim version: PLE

Re: Dependent Variable Noise

Post by bffcosta »

John,

I recommend you reading Appendix B of Business Dynamics (Sterman 2000) before modeling noise. I attached a possible structure you can use to model a first-order pink noise. Roughly speaking, pink noise is used when next value is in somehow dependent on previous one, it means that there is some 'inertia' or 'memory' in the process of generating noise.

I hope it helps.

Kind Regards,

Bruno.
Attachments
Possible structure to use pink noise.jpg
Possible structure to use pink noise.jpg (265.32 KiB) Viewed 3374 times
johnjohnson4
Member
Posts: 47
Joined: Mon Mar 14, 2016 4:58 pm
Vensim version: PLE+

Re: Dependent Variable Noise

Post by johnjohnson4 »

Thanks very much for your reply. It was very helpful!

V/r,
john
Post Reply