Question about value selection prediction algorithms?

Question

Algorithms

Good afternoon!
What methods/algorithms exist to predict value selections based on selection history analysis?
Given:
The organization has an accounting system, a group of users classifies transactions. That is, each document is assigned to a specific project. Documents may contain sets of parameters:
-legal entities
-contractors
-contracts -settlement
accounts
-individuals -operation item Depending on the operation, the set may be different, some parameters may be missing.
For the most part, the parameters are repeated in different versions, for example, one counterparty can work with different legal entities. persons or under different contracts depending on the project. The data is updated almost every day, I would like to help users with the choice of projects in new documents.
Challenge:
Automatically prompt the user for multiple (ideally one) project values ​​for each new document.
Now it is implemented in this way: We
analyze the history by a set of parameters (legal entity, legal entity - counterparty, legal entity - counterparty - contract, etc.) the more parameters matched, the greater the sample weight, sorted by weight, we get a list values ​​in descending order of the most likely value. Also, the number of repetitions of each option is not taken into account now.
Surely there are more advanced forecasting options for the described case. Tell me where to dig?

Answer 1

ANNs are quite suitable for forecasting tasks.