If you want to use a prediction model, then you need a well-defined target. In your case, the Utilization of an ATM is a vague term. You cannot measure it as it is right now. If you solve this problem, then what you are looking at is algorithms called Ranking Methods or Learning-to-Rank
Abstract from Wikipedia
Learning to rank or machine-learned ranking (MLR) is the
application of machine learning, typically supervised, semi-supervised
or reinforcement learning, in the construction of ranking models for
information retrieval systems. Training data consists of lists of
items with some partial order specified between items in each list.
This order is typically induced by giving a numerical or ordinal score
or a binary judgment (e.g. "relevant" or "not relevant") for each
item. The ranking model's purpose is to rank, i.e. produce a
permutation of items in new, unseen lists in a way which is "similar"
to rankings in the training data in some sense.
Let's move to another working domain to make the example easier to understand. The example was taken from TowardsDataScience tutorial page.
You have an e-commerce shop and you want to rank your products with that way that will be sorted on a search page and maximize your revenue. The features you have are the attributes of the product and the target is if the visitor bought or not on that session.
You train then a Classifier (LogisticRegression for instance) and get the probability of prediction for that class as the ranking.