non-optimization models

[CharlieG]

Hi everyone. Am new to group and have a question. I have used Yasai
a couple of times for business decision-making and like the
simplicity. Now, I have a 'non-optimization' problem, where I'm
trying to forecast demand for farm pesticides in a region. Demand is a
function of ... arable land acres, crop mix, pesticide pricing, and
one or two other variables. Based on past data, all of these variables
have ranges that resemble normal distributions with means/std.
deviations. The desired output is a histogram of the expected
Demand. Does anyone have any suggestions as to how to set up such a
model with Yasai? This is a little bit different than the
optimization examples portrayed on the website. Thanks, in advance.

[Jonathan Eckstein]

You would just model the random inputs with GENxxx functions, probably
GENNORMAL based on what you have described. Put the total demand
calculated in side SIMOUTPUT( . , "Demand"), and run this simulation.
You will only get one scenario, which is fine. YASAI should create a
sheet showing the mean, standard deviation, and percentile statistics
for demand. Then use YASAI Charts... to make the histogram.

It should look just like the examples, except without any PARAMETER cells.

One nitpick: "expected demand" normally means the expected value (mean)
of demand, which is a single number. Hence you cannot have a histogram
of it. But you can have a histogram of the whole sample of the random
variable, which one would ordinarily just call "demand".

-- Jonathan (benevolent dictator of YASAI)

[CharlieG]

Jonathan, my thanks for your quick response and instructions on the
non-optimization solution. I plan to try it out later this week with
some 'field' data. And, yes, fair nitpick on 'expected demand'. This
old geezer has always had a shaky grasp of statistics concepts and how
to use them. So bonus thanks for your tip - which I understand.
Stay cool.

[CharlieG]

Jonathan, thanks again for the hints for my demand forecasting
exercise. Ran it today with some old data and after working out a few
modeling bugs, it worked great. In addition to my own consulting
business, I am sometimes asked to accompany students from a certain b-
school on consulting projects. When simulation is appropriate, the
student teams have used their own 'black box' software like @Risk to
go about running their numbers. I see Yasai giving me the chance to
do a little quality control on the deliverable by double checking
their simulation assumptions and output. As this will involve a b-
school other than Rutgers, are you OK with this manner of my usage.
Regards.

[Jonathan Eckstein]

Of course -- it is open-source software and may be used for anything subject to the disclaimer that the authors have no legal liability for defects.

I designed YASAI in reaction to the many difficulties our students had with a long-ago version of @Risk.

-- Jonathan (on vacation)

Sent from my iPhone