The Honeymoon project is located 400 km northeast of Adelaide, approximately 75 km northwest of Broken Hill, within the Curnamona Province. The uranium deposit consists of five discrete mineralised sand packages, located near the confluence of a major tributary entering the Yarramba Paleochannel. The ore at Honeymoon exists in an underground aquifer and is extracted by in situ recovery (ISR), the chemical process of extracting minerals from the host rock underground through the utilisation of specially designed wellfields. Boss Energy and WGA (Wallbridge Gilbert Aztec) were granted an Accelerated Discovery Initiative (ADI) grant by the South Australian Government, to deliver this ‘proof of concept’ geophysical data processing tool for sedimentary uranium deposit evaluation for ISR during Greenfields exploration. The proposed Tool will take information available at the exploration stage of the project and predict ISR decline curve and uranium extraction. The Tool has the potential to assist operations in wellfield planning and integrate with process plant models for economic optimisation of uranium production. In this final report, WGA demonstrated a machine learning approach to the ‘proof of concept’ Tool, the development of which was supported by a thorough review of literature, Honeymoon historic operational datasets, and current modelling methodology. Key finding included: • Application of machine learning to predict decline curve is novel, • Current modelling techniques require a detailed profile of the deposit and require significant computing power, • Two proof of concept models were identified that were deemed suitable and, • That the selected proof of concept predictive model showed reasonable accuracy. The following opportunities have been identified, which have the potential to improve production planning and wellfield development: • The opportunity to utilise the Boss Infill Data Set, which includes data from the borehole magnetic resonance tool, density and neutron logs, which data may be used to further simplify input models. • Development of the decline predictive decline curve into a wellfield planning tool - sequence of well operation start-up can be optimised by overlaying the pattern decline curves and solving for target PLS grades and resource extractions. • Incorporating Kalman Filters: Once the simplified models are developed, there is potential for adaptive approaches such as Kalman filters to feedback measurements to forward predict production curves, which will further improve the accuracy of the model. • Application of machine learning to lithology interpretation.