Abstract: Certain aspects of the present disclosure provide techniques for determining a target recommendation for a transaction based on historical transaction data. An example technique for determining the target recommendation for a transaction based on historical data includes receiving a request for a target recommendation and transaction parameters from a user. Based on the transaction parameters, one or more subsets of historical transaction data are retrieved, and a respective weight applied to generate the target recommendation. Based on which of the one or more retrieved subsets of historical transaction data and respective weight(s) applied generate the target recommendation, a confidence level is generated. A determination is made whether the confidence level meets a minimum confidence level. Based on the determination that the confidence level meets the minimum confidence level, the target recommendation is provided.