b'Data trends Data trendsparsing and downloads extras (packages)electronic sensors, so let the algorithms allowing almost any data manipulation.paint their picture and critique their The Panda package seamlessly transformsthinking https://www.youtube.com/text and number columns for the NumPywatch?v=L_Guz73e6fw. Oz Minerals package to crunch matrix operations.found this approach useful to question Web services allows Python to call dataexploration bias due to classical deposit in from web sites on the \x1by so explorersmodels https://www.youtube.com/can compare their geological setting withwatch?v=uLqKODrPsUU.similar targets around the world.Why does elevation have the highest Machine Learning then assesses thecorrelation with occurrences in South relationship strength between all layersAustralia? Probably because there have with known occurrences (Figure1) andbeen nearly two centuries of exploration Tim Keepingpinpoints where more occurrences mightand fossicking in the close to home, well-Associate Editor for geophysical be (Figure2). populated Adelaide Geosyncline, instead data management and analysisof the hard-to-explore desert regions. technical-standards@aseg.org.au This is the oft maligned data drivenBut then, orogenies accumulate minerals version of machine learning (everythingand increase elevation. OK, can areas of in) as opposed to knowledge driven (usepast and present orogeny be visualised to chosen in\x1buences). But to paraphraseillustrate expected similarities? This is why Big data and machine learningSam Altman (CEO of OpenAI), engineersOz Minerals called the process a game of updates believe humans cannot see as much asgeological Battleships testing ideas.Some colleagues recently attended the PESA Python for Geoscience workshop https://pesa.com.au/events/pesa-qld-online-course-introduction-to-python-for-geoscience-2023/2023-08-15/. Sydney Universitys Nathaniel Butterworth showed the (relative) ease with which users can now analyse enormous amounts of data. While traditional exploration and GIS programs can display many layers simultaneously, trying to spot intersecting areas of interest may be di\x1ccult. Machine learning, and its dreaded data driven models, can cut some of that time involved.As we all juggle spreadsheets, a database or two, something for cross sections, a math programming language or specialist geochemistry plots, something to make a picture of drill holes in 3D and more and more, we all see the need to combine more aspects of geoscience.The number of datasets publicly available is extraordinary and growing. Universities, state surveys, GA and CSIRO pump out mountains of company, research and government collected data each year. NCI/AuScope/ARDC 2030 Geophysics Data Collection Project at ANU spearheads the \x1dle types required for large scale analysis https://ardc.edu.au/project/2030-geophysics-collections/. Python opens the door to accessing this data all at once.An appeal of the (free) Python high level programming language is that itFigure 1.Bar chart plot of most the signi\x1fcant input features associated with a mineral occurrence taken inherently deals with \x1dle handling and textfrom Sydney Informatics Hub Python for the Geoscience 2023 Python workshop.41 PREVIEW DECEMBER 2023'