Oil and Gas
Oil and Gas | Data Analytics
Applied Statistical Modeling and Big Data Analytics
This training course will provide a hands-on introduction to statistical modeling and big data analytics so that participants can use them for petroleum engineering and geoscience applications. Topics to be covered include: (a) easy-to-understand descriptions of the commonly-used techniques, (b) case studies demonstrating the applicability, limitations and value-added proposition for these methods, and (c) hands-on problems sessions using open source and/or commercial software. This course will provide engineers and geologists with practical techniques for identifying hidden patterns and relationships in large datasets and extracting data-driven insights towards actionable information that can contribute to lower cost, improved efficiency and/or increased productivity in oil and gas operations. This class will arm petroleum engineers and geoscientists with advanced capabilities to extract new insights from E&P data that can help: (a) learn hidden patterns and relationships in geologic datasets, (b) identify production sweet spots in developed plays; (c) determine factors responsible for separating good wells from poor producers wells, (d) build fast surrogate models of reservoir performance, and (e) assist in predictive maintenance by identifying failure inducing conditions from historical records.
Schedule
Duration and Training Method
This is a classroom or virtual classroom consisting of lectures interspersed with worked examples, hands-on exercises and discussions.
The textbook for the course will be “Applied Statistical Modeling and Data Analytics: A Practical Guide for the Petroleum Geosciences” by Srikanta Mishra and Akhil Datta-Gupta (Elsevier, 2017), supplemented by various technical publications.
Course Overview
Learning Outcomes
Participants will learn to:
- Apply foundational concepts in probability and statistics for basic data analysis
- Perform linear regression for building simple input-output models
- Conduct multivariate data reduction and clustering for finding sub-groups of data that have similar attributes
- Converse with confidence about big data, data analytics and machine learning terminology and fundamental concepts, and critically review topical technical publications
- Apply machine learning techniques for regression and classification for developing data-driven input-output models
- Evaluate proxy modeling and uncertainty quantification studies for probabilistic performance forecasting
Course Content
(1) Foundational Concepts
- Big data technologies, basic data analytics and machine learning terminology/concepts
- Data, statistics, and probability
- Distributions (models, fitting distributions to data)
- Inference (Confidence limits, bootstrap, significance tests)
- Data visualization
- Problem session
(2) Basic Regression Analysis
- Linear regression (univariate and multivariate regression)
- Understanding regression statistics, ANOVA
- Non-parametric regression
- Problem session
(3) Multivariate Data Analysis
- Dimension reduction (Principal component analysis)
- Cluster analysis (K-means, hierarchical clustering, self-organizing maps)
- Problem session
(4) Machine Learning Basics
- Overview of techniques
- Evaluating model performance (model validation, goodness-of-fit, common pitfalls)
- Variable importance
- Model aggregation
- Case studies
(5) Machine Learning for Regression and Classification I
- Classification/regression trees
- Random forest
- Gradient boosting machine
- Problem session
(6) Machine Learning for Regression and Classification II
- Support vector machine
- Neural networks and deep learning
- Problem session
(7) Miscellaneous topics and Wrap-up
- Experimental design and response surface analysis
- Uncertainty quantification
- Selected literature review
- Key takeaways and resources
- Data analytics do’s and don’t’s
Who Should Attend and Prerequisites
This course is for designed for petroleum engineers, geoscientists, and managers interested in becoming smart users of statistical modeling and data analytics.
Instructors
Srikanta Mishra
Background
Dr. Srikanta Mishra is Technical Director for Geo-energy Modeling & Analytics at Battelle Memorial Institute, the world’s largest independent contract R&D organization. He is responsible for leading a technology portfolio related to computational modeling and data analytics for geological carbon storage, improved oil recovery projects, and shale gas/oil development. His recent work has focused on full-physics and reduced-order modeling, and pressure-based monitoring, of CO2 geologic sequestration projects. He has served as PI or co-PI on a number of CO2 geological storage and EOR projects funded by the US Department of Energy with field demonstration sites in the Appalachian and Michigan Basins.
Dr. Mishra has presented lectures and conducted short courses and workshops on CO2 geologic sequestration in many US universities as well as in academic and research organizations in Switzerland, India, South Africa, UK, Mexico and Indonesia. He is an editor of the book “CO2 Injection in the Network of Carbonate Fractures” recently published by Springer, and the author of ~200 technical publications.
Dr. Mishra is a member of the Technical Advisory Board of the SMART initiative (Science Informed Machine Learning for Accelerating Real-time Decisions for Subsurface applications) that is organized by the US Department of Energy’s Carbon Storage Program and involves multiple nationals labs and universities. He was selected as an SPE Distinguished Lecturer for 2018-19 on the topic of Big Data Analytics. He has also served as an Adjunct Professor of Petroleum and Geosystems Engineering at The University of Texas at Austin. Dr. Mishra holds a PhD degree from Stanford University, an MS degree from University of Texas and a BTech degree from Indian School of Mines – all in Petroleum Engineering.
Affiliations and Accreditation
PhD Stanford University - Petroleum Engineering
MS Stanford University - Petroleum Engineering
BTech Indian School of Mines - Petroleum Engineering
Courses Taught
N479: Applied Statistical Modeling and Big Data Analytics
N480: Introduction to Statistical Modeling and Big Data Analytics
N535: Carbon Capture Sequestration (CSS)
N567: Carbon Capture, Utilization and Storage