Introduction to Subsurface Machine Learning
INSTRUCTOR: Michael J. Pyrcz, PhD
DISCIPLINE: Geoscience, Engineering, Unconventional Reservoirs, Multi-Disciplinary & Introductory
COURSE LENGTH (DAYS): 2 Days (Classroom) / 4 Half-Day Sessions (Live Online)
AVAILABILITY: Public & In-House
ATTEND AN UPCOMING CLASS:
WHO SHOULD ATTEND: Technical energy industry professionals (petroleum engineers, geoscientists) with basic Python proficiency.
COURSE DESCRIPTION: This two-day workshop focuses on the advanced application of data analytics, geostatistics, and machine learning to energy industry data. The course is a critical step in laying the foundation necessary for thinking statistically and identifying the key signals from the noise that is data.
Specifically, this workshop will teach participants:
- To effectively prepare data for deep dives with advanced analytic techniques and ensure that any drawn conclusions are trustworthy and reliable.
- To glean insights and make predictions from your data, using techniques such as outlier detection, data debiasing and imputation, feature engineering, anomaly detection, supervised and unsupervised learning, spatiotemporal modeling, and uncertainty modeling.
- To understand the assumptions and limits of data precision, scale and coverage, spatial interpolation, multivariate models, analytics and uncertainty models, given that predictions are only as strong as your process.
- What Big Data really means and how to apply robust machine learning tools to real Energy data problems (Dask, Torch, TensorFlow etc).
- Primary Goal: Introduce subsurface data analytics, geostatistics and machine learning methodologies, workflows, and applications
- Feature Ranking, Dimensional Reduction, Naïve Bayes, Clustering, K-Nearest Neighbors, Decision Trees, Ensemble Trees, Support Vector Machines, Neural Networks
- Data preparation methodologies
- Introductory Geostatistics.
SAMPLE TOPICS FROM THE CLASS:
"Subsurface Machine Learning: Introduction to Spatial Data Analytics with Python"