Forecast

[Jindu]

Hi,
I’m trying to forecast the future for example picking the last value of the historical order cycle (in this case 34 being the last value of the historical order pattern) then use it as the start value of the forecast to see the future order pattern. Is there any way I can do this? An example of the structure have been attached.
Many thanks and kind regards.

[Administrator]

I don't understand what you are asking. The model you've uploaded is using the forecast function, is it not doing what you need?

[Jindu]

Thanks for your reply. What I’m trying to do is to project the future customer order from where the historical actual order cycle ends. For example, if the historical is 60 weeks and the last value for the 60th week is 34 units/week. I want to then project or forecast for example from the 60th week (34 units/week) to 90th week from the historical order. The model uploaded with the forecast function still start from the same old value I don't know if you understand.

[Administrator]

I'm still struggling to understand what you are trying to do.

Does this help?
https://www.vensim.com/documentation/in ... ckward.htm

[Jindu]

Thanks.

[tomfid]

In your FORECAST(Actual Order Cycle, 10 , 20 ) equation, the 10 and 20 should be made into named parameters, both for dimensional consistency and to expose them to sensitivity testing and make the model structure more portable.

[tomfid]

I think you want something like the following:

historic order rate = order cycle(time)
last historic order rate = order cycle(switch time)
future order trend = ... {this could be a constant, or estimated from the lookup}
future order rate = last historic order rate + future order trend*MAX(0,time-switch time)
switch time = 60
order rate = IF THEN ELSE( time <= switch time, historic order rate, future order rate )

[Jindu]

Many thanks for your suggestion the order rate works (projects) as I intended.

[sankee82]

Hi!
I am also trying to forecast the future variables by using the historic data. In my case my variable of interest is modeled as a level whereas In the above case the order cycle was a lookup. Can you please suggest me how to use the above algorithm for a level/stock i.e., Auxiliary variable?


Thanks and Regards,
Sankeerthana.