PLR - Engineering Solutions & Software
16
Model Explainability - Simulation
-
Simulation example
Example Use Cases
Selected Integrity and Risk Management Solutions for Pipeline Systems
ML.ai platform supports any integrity or risk use case, leveraging purposed processes & your existing data investments
Best Viewed at 1040p
TPD
Third Party Damage Susceptibility
-
Learn Classification Model to Predict Third Party Damage Susceptibility
-
Assess Data Quality Metrics
-
Measure Contribution of Predictors & Simulate PMM's
ILI
Prioritize In-Line Inspections
-
Learn & Validate Model Based on Past Inspections
-
Apply Model to Similar Pipelines
-
Use Box-Plots to Prioritize Pipelines & Identify Outliers
SHA
Model Explainability
-
Basic Example Demonstrating Shapley
-
Method Deconstructs any Output Result Set
-
Results are Explainable, Human Readable and Actionable
MLP
What is a Typical Machine Learning Process
-
Establish prediction objectives
-
Create learning & test data
-
Learn model and measure performance
-
Make predictions, support use cases, perform simulations & influence analysis
SHA
Model Explainability
-
Basic Example Demonstrating Shapley
-
Method Deconstructs any Output Result Set
-
Results are Explainable, Human Readable and Actionable
MT
Machine Learning Types
-
Supervised & Unsupervised
-
Introduction to Classification & Regression
-
When to Use Each
ME
What is a Machine Learning Method
-
Classification & regression
-
100's of methods to choose
-
Selection criteria - performance, cost, explainability, preference
CV
Cross Validation
-
Cross-validation
-
Testing
-
Learning Curves
CP
How to Measure Classification Performance
-
Metrics
-
Confusion matrix
-
Learning curve
RP
How to Measure Regression Performance
-
Metrics
-
Unity plot
-
Learning curve
Leverage Your Existing Data Investments
One process and one platform supports any integrity or risk use case, all with the intent on leveraging your existing data investments