Skip to content
AfriScienceGroup home
Petroleum Geoscience

Machine Learning and Data Science for Petroleum Geoscience

Geoscience departments hold decades of logs, cores, seismic and reports that are rarely analysed together because the data are messy and spread across formats. Data science can make that archive productive when the geoscientist understands the algorithm. Delegates start with Python, notebooks and the core libraries, then clean and condition well logs, handling gaps and depth shifts. Supervised learning follows: regression for synthetic logs and permeability prediction, classification for facies, with honest blind well validation. Unsupervised clustering for electrofacies and dimensionality reduction come next, then convolutional networks for fault and salt detection. The final sessions treat explainability, bias and how to deploy a model into a team workflow. Delegates leave able to build, test and critique machine learning models on their own data.

5 day course

Course objectives

  • Prepare and clean well log and core data sets in Python
  • Apply regression methods to predict missing logs and permeability
  • Train and validate classification models for facies prediction
  • Use clustering and dimensionality reduction to define electrofacies
  • Explain how convolutional neural networks are applied to seismic interpretation
  • Evaluate model performance with cross validation and blind wells
  • Interpret model outputs with explainability tools and geological judgement
  • Plan the deployment of a model into an existing geoscience workflow

Who should attend

  • Geoscientists with an interest in data analytics
  • Petrophysicists automating interpretation
  • Seismic interpreters exploring automated picking
  • Data scientists joining subsurface teams
  • Digital transformation leads in upstream companies

Course outline

  1. 01Python and notebooks for geoscientists
  2. 02Subsurface data wrangling
  3. 03Exploratory analysis and visualisation
  4. 04Supervised regression
  5. 05Facies classification
  6. 06Clustering and electrofacies
  7. 07Deep learning for seismic
  8. 08Validation, explainability and bias
  9. 09Deploying models in the team

Scheduled sessions

Scheduled sessions for Machine Learning and Data Science for Petroleum Geoscience
DatesVenueFormatPriceRegister
26 to 30 October 2026Kampala, UgandaClassroomUSD 1,325 per delegate
9 to 13 November 2026Kigali, RwandaClassroomUSD 1,325 per delegate
7 to 11 December 2026OnlineOnlineR8,500 per delegate
4 to 8 January 2027Windhoek, NamibiaClassroomR19,950 per delegate
15 to 19 February 2027Lagos, NigeriaClassroomUSD 1,325 per delegate

Machine Learning and Data Science for Petroleum Geoscience

From R8,500

Ask