Page 1 of 1

empirical distribution

Posted: Friday 01 April, 2011 - 11:56
by tokkitrooi
Hello,

How can I integrate a empirical distrubtion in the right way?

In the empirical distribution you can give the atoms some values, so for 10% ''40'' en for 30% "50" and else "80"

so the table will be:

10 40
40 50
100 80

But, has the emp. distribution be connected to the source? and how can i tell the server that he must take the value of the em. distribution? in the cycletime, i can not say something like : Label( blabla), because i don't make any labels...

thank you!

Re: empirical distribution

Posted: Friday 01 April, 2011 - 12:13
by MatthijsJongboer
When using the empirical distribution atom, you can choose a distribution name. This name becomes a global 4DScript that can be used throughout the model without having to connect any channels.
Attached the example model from ED. I revised it so the server now takes the values from a new Empirical Distribution according the data you provided.

Re: empirical distribution

Posted: Friday 01 April, 2011 - 15:15
by tokkitrooi
Ok, thank you! but the following is the case:

If the arrival time is for example 10 (and that happens in 10% of the cases) , the cycle time has to be 50 or something like that.

The file has only to empircal distrbution with no link ( i think):

because now it is possible that an arrival of 10 can have a cylcletime of 100 or something like that. How can you model this?

Re: empirical distribution

Posted: Tuesday 05 April, 2011 - 13:47
by MatthijsJongboer
The arrival time is something different. In the model, the cycletime is chosen based upon the provided emp.distr.
I understand form your remark that there is a relation from the arrivaltime and processtime?