AI algorithm for adjust percentages and constant coefficients values

I’m making an application that will help the people to donate stuff in one delivery point and then the application will determine who suits the item donated perfectly (the beneficiaries who receive the donations are inserted in the database previously)

I’m making an algorithm to determine the suitable beneficiary from those in the database based on parameters such need scale, distance, last receiving date, etc.

I gave the distance the priority of 50% and the need scale 30% and the last receiving date 20% as the percentages of the parameter that the algorithm will use to determine the suitable beneficiary.

But am sure that the priority percentages I gave are not accurate, so I decided to create an ai model to determine these percentages.

I’m using python tensorflow to create a model needed. And now i’m looking for a supervised learning algorithm that suits my situation. I just need the model to return a set of the new percentages based on the model supervisor’s opinion which is really close to reinforcement learning algorithms too. I created a csv file to save all the training data needed to feed the model.

Any ideas?
I’m looking for the optimal algorithm for this situation.