question

mike S4 avatar image
1 Like"
mike S4 asked Brenton King commented

How to predict the inter-arrival distribution of a source?

Hello everyone,

I need to vary the value associated with a label of each item in the warehouse based on the demand for arrival.

In my model I have a source with a known inter-arrival distribution, while the warehouse has items with a label that should be increased according to the arrivals: the more frequent these are, the more it increases and the less they are frequent and the less it increases.

Knowing the inter-arrival distribution of the source, is there a way, from the point of view of the warehouse, to estimate or predict the arrival distribution of the source and thus modify the label?

Thanks for the help.

FlexSim 19.0.0
sourcedistributionsinter-arrival time
· 6
5 |100000
Toggle Comment visibility. Current Visibility: Viewable by all users

Up to 12 attachments (including images) can be used with a maximum of 23.8 MiB each and 47.7 MiB total.

Joerg Vogel avatar image Joerg Vogel commented ·

How long do you want a forecast into the future?
You can create from the statistical distribution several values for events in the future and store them in an array. When a creation event occurs, you return in interarrival time the first element of the array, destroy the element and append at the end a new value. You can get the next events from the array to estimate a forecast.

1 Like 1 ·
mike S4 avatar image mike S4 Joerg Vogel commented ·

@Jörg Vogel thanks for the answer. In my model I have to predict the daily arrival of customers in a shopping center, knowing the distribution of arrival. So, for example, if more customers arrive around 9:00 a.m., the labels' values of the items in the warehouse should increase rapidly during that period and accordingly if fewer customers arrive, the labels' values should grow more slowly. I need to know in advance how many customers are coming in that time. How can I implement this behavior into FlexSim?

0 Likes 0 ·
Braydn T avatar image Braydn T mike S4 commented ·

@mike S4

We are having trouble determining exactly what you are supposed to do. If you are trying to create items in a warehouse depending on the number of people that arrive, you can just use an event triggered source for a persons arrival and increase that way. You can't know the number of people that will be in a place at a given time, which is why you use a statistical distribution so you can get an idea of something that would simulate reality. Could you upload an example model of something similar to what you are trying to accomplish?

1 Like 1 ·
Show more comments

1 Answer

Joerg Vogel avatar image
2 Likes"
Joerg Vogel answered Brenton King commented

You can't predict any random values. You can create a list or array of random values. Whenever You pull a value from the list, you push simultaneously a new value to the list. Because you know the entries of the list you can estimate the next events in the future.

A User Command evaluates a Global Variable string containing a statistical distribution and create an array of 10 values. The user command pulls from the global variable array the first value and push a new value to the end. 2 model triggers fill the array OnModelOpen and OnModelReset with 10 values hard coded.

initial-rand-values-global-variables.fsm


initial-rand-values-global-variables.fsm (18.1 KiB)
· 1
5 |100000
Toggle Comment visibility. Current Visibility: Viewable by all users

Up to 12 attachments (including images) can be used with a maximum of 23.8 MiB each and 47.7 MiB total.

Brenton King avatar image Brenton King ♦ commented ·

I have had a few models where I needed to "predict" what was going to occur. Jorg is correct, you can't truly predict what the next random value will be. In my model, I kept track of the last X number of arrivals and would periodically run my prediction for the future based off of the last arrivals.

Another method, would be to have all of your arrivals created by the source and then sent through an initial delay process. The initial delay would allow you to see that they are coming and prepare accordingly with a "prediction". This is kind of cheating because you aren't really predicting anything, you are looking into the future. All of your arrivals would be delayed by the same initial time period so it shouldn't affect your stats.

Just my quick thoughts.

0 Likes 0 ·