Modules

2Data Infrastructure & Architecture

3Digital Workflows in E&P

4Well Surveillance Tools

5Production Optimisation

6Dashboards & Data Analytics

7AI & Machine Learning

8Digital Twins & Simulation

9Automation & Remote Operations

10Cybersecurity in OT/SCADA

11Change Management & Digital Culture

12Course Exam

Module 7•Unit 5 of 5•5 min

Knowledge Check

1. Which type of ML learns from labelled input-output pairs?

Unsupervised learning Supervised learning Reinforcement learning Transfer learning

2. What is the main advantage of a virtual flow meter over a physical multiphase meter?

Higher accuracy in all conditions Much lower cost and continuous monitoring without dedicated hardware It requires no sensor data It eliminates the need for any well testing

3. What is model drift?

A model that runs too slowly When a model is shared between teams When changing data characteristics cause model predictions to become less accurate over time A model trained on too much data

4. Which algorithm family is most popular for tabular production data in oil and gas?

Convolutional Neural Networks Tree-based ensembles (Random Forest, XGBoost) Generative Adversarial Networks Support Vector Machines

5. What does MLOps stand for?

Machine Logic Operations Machine Learning Operations Model Lifecycle Optimisation Process Multi-Layer Output Processing

6. What percentage of project effort should be allocated to MLOps according to best practice?

Less than 5% 10% 20-30% Over 80%

Model Deployment & MLOps

Module 8: Digital Twins & Simulation