how to calculate the standard error in vensim

sally · Post by **sally** » Tue Nov 14, 2006 10:43 am

i want to calculate the demand's STD,

can some one tell me how to do it?

does it have any function in vensim can use to calculate one variable's STD

thank you ....

bob@vensim.com · Post by **bob@vensim.com** » Tue Nov 14, 2006 1:38 pm

the stats tool will report the standard deviation of a variable over a simulation.

sally · Post by **sally** » Tue Nov 14, 2006 1:50 pm

i don't understand your meaning.

can you explain clear to me

thakn you ...

bob@vensim.com · Post by **bob@vensim.com** » Tue Nov 14, 2006 10:16 pm

Chapter 14 of the Refrence Manual describes the Stats tool.

sally · Post by **sally** » Wed Nov 15, 2006 6:04 am

thank you !

but if want to collect at t=1 STD, then use this number to calculate something

how to do it

bob@vensim.com · Post by **bob@vensim.com** » Wed Nov 15, 2006 10:56 am

You can only see current values in model equations. To get a standard deviation over time requires multiple time values. You may be able to invent some equations that compute this for past time - if you do post them here as other people may also be interested.

LAUJJL · Post by **LAUJJL** » Wed Nov 15, 2006 12:42 pm

Hi

Joined a model where you have every time steps, a set of variables generated randomly but can be loaded from real data, and that calculates each time step, the mean and the standard deviation of the set of variables. Vensim pro needed.
Regards.
Jean-Jacques Laublé

sally · Post by **sally** » Wed Mar 14, 2007 2:41 am

hi,

if the experience is not a set, just a constant or a uumber, can i use this model?

because i want to calculate the demand's STD, and every week's demand is different, i want to get the STD, then to calculate the ROP.

hope you know my means,

thank you .

LAUJJL · Post by **LAUJJL** » Wed Mar 14, 2007 6:53 am

Hi

You need to be more explicit about your needs.

The standard deviation is not necessarily the solution to your problem.

If you want the std from a variable, you have to define a trend for that variable, wich is for every time step, an expected value for that variable base on its historical data, and then calculate the standard deviation of the real value from that expected value.
The simpler solution and generally the more efficient to predict the demand is to take a running average eventually
weighted with the oldness of the data.
Did you study the afore mentioned chapters from business dynamic's?
Regards.
JJ