The most used Machine Learning algorithms in Petroleum Engineering

Alfonso R. Reyes
(28 March 2018)


Continuing with the article on machine learning (ML), from the papers extracted in OnePetro, I have been able to detect the use of these ML algorithms:

  • Neural Network

  • Genetic Algorithm

  • Support Vector Machine

  • Principal Component Analysis

  • Linear Regression

  • Fuzzy Logic

  • Hierarchical

  • K-Means

  • Singular Value Decomposition

  • Decision Tree

  • Support Vector Regression

  • Deep Learning

  • Logistic Regression

  • Boosting

  • Random Forest

  • Nearest Neighbor

  • Discriminant Analysis

  • Gaussian Mixture

  • Gaussian Process Regression

  • Naive Bayes

  • Hierarchical Clustering

  • Reinforcement Learning

  • K-Nearest Neighbor

  • Hidden Markov

  • C-Means

  • Fuzzy C-Means

  • Gaussian Mixture Model

  • Kernel Density Estimation

  • Gradient Boosting Tree

  • Kernel Approximation

With a little bit of R magic:

From all these techniques, the top 20 most used are:

If you see I am forgetting a specific machine learning algorithm, please let me know.

Notes. (1) Fuzzy-Logic is not considered a ML technique; rather, belongs to control theory. But still, I am including it here because there is a considerable number of papers using the algorithm. (2) I grouped Convolutional Neural Networks with “Deep Learning”.

So, this is the state of use of machine learning in petroleum engineering.