What do we assume for volumes?

N-Wouda / Urban-Waste-Collection

Code for our paper "An integrated selection and routing policy for urban waste collection".

GNU Affero General Public License v3.0

2 stars 0 forks source link

What do we assume for volumes? #57

Closed N-Wouda closed 6 months ago

N-Wouda commented 1 year ago

Right now that's uniform as $U[30, 65]$, for each container. I think the following two scenarios make perhaps more sense:

Triangular, between [30, 65], with the mode at 47.5.
Same as above, but mode is sampled from $U[40, 55]$ for each container. This relaxes the assumption that these distributions are the same for each container.

N-Wouda commented 1 year ago

@ndvanforeest what do you think? I talked to Umur about this and he thinks we especially need to do something about the iid assumption.

ndvanforeest commented 1 year ago

The Idd assumptions seems the only sensible options. In other words, if you see me with a big bag, would that imply that the next customer has a small, or a big bag? All in all it does not matter the much. The sum of 80 uniform rvs, whether identical or not, is extremely well modeled as a normal distribution. Sent from my iPadOn 9 Sep 2023, at 13:55, Niels Wouda @.***> wrote: @ndvanforeest what do you think? I talked to Umur about this and he thinks we especially need to do something about the iid assumption.

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: @.***>

N-Wouda commented 1 year ago

That's not what I asked?

N-Wouda commented 1 year ago

Ah, I see where this went wrong. I called it iid, but I meant to say that we currently assume the same volume distribution for each container. That makes very little sense to me, but it was convenient to work with. I discussed this a bit with Umur yesterday, and we think we should make this a bit more random.

ndvanforeest commented 1 year ago

I see now. You mean that the volume distribution should not be the same for all containers. Sure, that makes sense. However, setting up a good sampling strategy might not be completely straightforward.Sent from my iPadOn 9 Sep 2023, at 21:28, Niels Wouda @.***> wrote: Ah, I see where this went wrong. I called it iid, but I meant to say that we currently assume the same volume distribution for each container. That makes very little sense to me, but it was convenient to work with. I discussed this a bit with Umur yesterday, and we think we should make this a bit more random.

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: @.***>

N-Wouda commented 6 months ago

OK, I don't think we need to do this for the paper. I doubt it really influences the results because the overflow model should be able to pick these differences up quickly anyway.