January 7, 2019 - January 8, 2019
8:00 am - 12:00 pm
Learn the basics of Machine Learning thru practical examples in pipeline integrity and introduction to open source software. This course will demonstrate the fundamentals of performing Machine Learning, so the participant will have the know-how to apply these to their business.
- Machine Learning reveals useful patterns hidden within complex data streams for the purposes of understanding what is likely to happen in the future,
- Understanding these patterns supports identification of higher risk or threat susceptible assets subsequently enabling more efficient intervention and proactive mitigation,
- Machine Learning improves the efficiency of data management activities by predicting the most influential variables impacting integrity objectives,
- As a data driven algorithmic approach, Machine Learning may be used to improve or replace existing risk analysis applications
Students are required to bring their laptop to this training
Who Should Attend?
–Integrity Managers & Engineers
–Risk Managers & Engineers
This 1½ day class provides instruction on the practical application of Machine Learning to integrity management of pipelines. The class will demonstrate how Machine Learning is used to determine threat susceptibility based on historical data and domain expertise, the results of which are used to support overall risk management and optimal spend decision-making. Machine Learning is a collection of methods and processes to help understand what is likely to happen in the future based on what has happened in the past. Practical examples will be presented to demonstrate the process, value and challenges of Machine Learning and the participant will be provided an intuitive open source data science platform with spreadsheet examples.
The attendee will:
- Learn the basics and process of using Machine Learning
- Learn how to establish prediction targets (corrosion, cracking, third party, etc.)
- Learn how to select required data (dependent, independent, correlation, etc.)
- Learn how to prepare data (relational, spatial, dynseg, aggregation, missing data, etc.)
- Learn how to select a learning model (classification, regression, ensemble, etc.)
- Learn how to measure model performance (training data, validation data, confusion matrix, etc.)
- Learn how to optimize a model (selecting the best parameters for the learned model)
- Learn how to score data (applying the model against data to make predictions)
- Understand model deployment (integrating the model in the enterprise)
–Introduction & Course Objectives
–Value of Machine Learning to Pipeline Integrity
–Machine Learning Process
–Algorithms & Operators
–Case Study – External Corrosion
–Case Study – Third Party
–Extending Results to Risk Management
–Introduction to Open Source Machine Learning Platform
“Great visuals–will be using as a resource in the future.”
“The technical scope was very thorough. Mike is an excellent resource.”
“I’m excited to further explore this topic and determine ways to implement in the ‘real world.'”
“I liked seeing live troubleshooting.”
Michael Gloven is managing partner of EIS, a provider of life-cycle based risk engineering solutions supporting the optimization of asset condition assessment, rehabilitation and replacement programs in the water, energy and renewable infrastructure industries.
Mike has more than 30 years’ experience in the pipeline industry and started his career with Conoco before co-founding and holding the role of president of American Innovations (Bass-Trigon) and NRG TECH. Both companies provide technology based engineering solutions to the hazardous liquids and natural gas industries.