Automated ML.NET Training, Metrics, and Accuracy

Course details

Knowing how to use Microsoft ML.NET is one of the most in-demand skills in today’s job market. If you’re looking to boost your know-how with machine learning, ML.NET is a must-have in your toolbox. In this course, senior software engineer and Microsoft MVP Sam Nasr provides a comprehensive overview of some of the advanced features of ML.NET, including how to use the command line interface, gather metrics, and evaluate and improve model accuracy and performance. Get started with an introduction to automated machine learning (AutoML) before learning how to automate model training with the ML.NET command-line interface. Along the way, Sam shows you how to evaluate an ML.NET model with metrics and improve model accuracy with cross-validation, hyperparameter tuning, pipeline value inspection, and more.

Learning objectives
Build a model using automated machine learning (AutoML).
Evaluate the model.
Collect metrics on model performance.
Improve model accuracy.

Concepts

Introduction

Leveraging automation, metrics, and accuracy in ML.NET
What you should know
Using the exercise files

Using AutoML

What is AutoML
How to use AutoML API
Demo - AutoML

Automate Model Training with the ML.NET CLI

How to install the ML.NET CLI tool
Demo - Analyze sentiment using the ML.NET CLI
Telemetry in ML.NET

Evaluate the ML.NET Model with Metrics

Model metrics
Demo - Metrics
Permutation feature importance (PFI)
Demo - Permutation feature importance (PFI)

Improve Model Accuracy

Reframe the problem
Provide better data samples
Cross-validation
Demo - Cross-validation
Hyperparameter tuning
Inspect pipeline values
Demo - Inspect pipeline values
Choose a different algorithm

Conclusion

CLI pitfalls
Next steps