In 2020, under a State Government funded program to test the application of machine learning (ML) to government data, the Dept for Energy and Mining (DEM) and the Australian Institute of Machine Learning (AIML) ran a trial to test the capability of ML to predict rock outcrop across the state using a variety of different remote sensing datasets and existing mapping data from DEM’s geological mapping programs. The AIML developed a tool that works by taking locations of known, mapped outcrops (known as the ground truth) and remote sensing imagery and uses machine learning, specifically deep learning, to train a model that can predict the location of the ground truth from the remote sensing imagery. The tool can then take the same remote sensing imagery for the state of South Australia and predict the probable locations of outcrops across the entire state and also output a value that indicates the confidence in its predictions. The Automated Outcrop Prediction Package is the outcomes from this project. This file contains the results of running the outcrop prediction tool using the following datasets: orthoimagery (WMTS); radiometrics data; night time thermal (NTT) imagery; Shuttle Radar Topographic Mission digital terrain data (DTM); and valley bottom flatness (VBF) data.