Q Regarding the Medians, Std Dev for Stridor...

[RyanCrierie]

I'm speccing out the XML files for the following aircraft for my Aircraft Improvement project (which may include MORE heavy artillery as a bonus).

So far, the pack will include the following 3d shadows:

Axis:
Bf-109
Bf-110
Ju-88A-4 (and P-1, P-2 and P-4)
Fw-190
Hs-123
Hs-129

Allied:
P-39
P-40
Hurricane
Pe-2

Now, I'm guessing that if you enter just the starting dates and end dates; you get a nice slope going like:



with the median high point being about halfway between start and end.

I'm assuming that the "median service" date is a way to shift the curve back or forwards, to help simulate the sudden starts or stops of a war:

For example, the Il-10 Sturmovik; it first entered operational service in January 1945; but the war in Europe ended in May 1945; so that would "mess up" the median curve if we tried a start date of Jan 45, and a End Date of May 1945; it would have the median high point being somewhere between Jan 45 and May 45.

I'm guessing that if you put in "May 1945" as the Median High Point, it would instead show a slow start in January 1945, building up to the high point of May 1945?

And I'm guessing that "Median Prevalence" is how "high" the curve is; e.g a very popular unit like the Panzer IV would have a median prevalence of around 100% -- it was the most common tank in the panzer arm at the end of WW2; while something a bit more unique, like a King Tiger, would have a prevalence of only 25%?

[spellir74]

I'm interested in this too.

Specifically, does a stddev of say 12 mean 12 months on either side of the 'medserv' or 6 months on either side? And I assume that 'month' starts at the end of the month not the beginning. (I could explain that previous sentence better --but I can't hear myself think!)

[Stridor]

Ryan,

For a *true* normal curve all you need define is the median, the median prevalence and the standard deviation.

However the RBG adds 2 futher fields: start date and end date.

In terms of the normal curve, the smallest quantum is one month.

The terms

Start Date: ("month", "year") = The first date the unit became available *on the battlefield*. This field is MANDATORY - get from historical records
End Date:("endservice") = The last date the unit was ever seen *on the battlefield*. Defaults to 6/45 if omitted
Median Service:("medservice") = The date when the unit was at its *maximum* on the battlefield. Defaults to Start Date + 1 year if omitted.
Median Prevalence:("medprev") = The relative prevalence of the unit at its median service date "medservice". Defaults to 5 if omitted
- This value will depened on other medprev values for units in the same category (ie armour, transport, innfantry, air, etc)
- you could make this a % value if you knew the breakdown. Eg PzIV(X) 6/44 = 80% of entire tank force.
Standard Deviation:("stdev") = one standard deviation in months. Defaults to 12 if omitted.
- Eg. if set to 12 => 6 months before and 6 months after the median prev date 68% (1sd) of the entire battlefield presence was encounted.

Using these numbers you can do all sorts of interesting things, including skew normals. This system was choose as it contained the most amount of prevalence data in the smallest number of variables.

Erik was responsible for setting the stock values.

Try some values for yourself. After you generate a random battle look at the RBG_ZD.txt file in the root pck directory and it will give you a full statistical breakdown of the units it considered along with their % chance to appear.

So far this system along with the RBG has generated resonable random unit statistics.

Regards

S.