Hey everyone! I'm currently participating in a hackathon where we need to assign 20,000 orders to 200 couriers. Initially, we considered using a neural network to find various routes, but we've realized that with just two weeks left, that might not be feasible. Instead, we're thinking about creating a hybrid machine learning model combined with an algorithm. We thought about using a Mixed-Integer Linear Programming (MILP) approach, but we feel it might be too greedy. What other algorithms could you recommend for efficient routing?
2 Answers
With the number of parameters you mentioned, you might be able to apply a Support Vector Machine (SVM) to tackle your problem effectively. It's a straightforward approach that could fit your needs well!
Have you considered using a Dijkstra variant? I'm curious about how a neural network would fit into your plan. What exactly would you train it on? If you have a distance matrix, maybe unsupervised learning or clustering might be worth exploring.
We have a distance matrix, like {"from":20258,"to":1337,"dist":3492}, so Dijkstra isn't suitable for our case.