Machine-Learning

Machine learning techniques applied to subsurface data.

This repository contains notebooks on various unsupervised and supervised machine learning techniques applied to subsurface geoscience data. Machine Learning (ML) and Data Analytics (DA) have a wide range of applications in geosciences and are powerful tools for optimizing geophysical and geological evaluation workflows. There are several clustering, classification, and regression algorithms applicable to subsurface data analysis. However, an understanding of these techniques is required to develop fit-for-purpose methodologies for subsurface characterization. In addition, a key consideration when applying ML and DA to subsurface datasets is the choice of features or attributes that can enhance the interpretation workflow. These notebooks cover process-based workflows ranging from data handling and preprocessing to quantitative data analysis for common geoscience applications.

Contents

1. Unsupervised Seismic Facies Clustering (Statistical Methods)
2. Unsupervised Seismic Facies Clustering (Neural Networks)
3. PVT Bubblepoint Pressure Prediction (Regression - Statistical Methods)
4. Well Log Porosity Prediction (Regression - Statistical Methods)
5. Shell (In)Direct from the Source Challenge (Classification - Statistical Methods)
6. AERA Steam Optimization and Other Oddities Challenge (Regression - Statistical Methods)

Upcoming

1. Supervised Well Property Regression (Statistical Methods)
2. Supervised Well Property Regression (Neural Networks)
3. Supervised Well Facies Classification (Statistical Methods)
4. Supervised Well Facies Classification (Neural Networks)

Feedback

Suggestions and comments are welcome and appreciated.