Systems and methods of optimizing resource allocation using machine learning and predictive control

Abstract

A computer system includes a transceiver that receives over a data communications network different input data sets from one or more source computers communicating with the data communications network, where an input data set includes data objects, each data object including associated data object attributes. A processing system processes the input data sets using a predictive machine learning model to predict for a predetermined time period a list of predicted object attribute values for data objects in the input data sets. The list of predicted object attribute values is sorted to generate a current, ranked list of data objects with predicted data object attribute values, which may be modified to account for a prior ranking of data objects. A subset of lower ranked data objects from the prior ranking of data objects is replaced with a subset of higher ranked data objects from the modified, ranked list of data objects with predicted object attribute values to generate a new ranking of data objects. Performance metric(s) for data objects are calculated for data objects in the new ranking of data objects relative to benchmark data for the data objects in the new ranking of data. The predictive machine learning model may be retrained based on the performance metric(s) to improve its performance.