Data Professional’s Guide to Automated and Explainable Machine Learning

An increasing number of open-source and commercial vendors are producing tools that automate many tasks of developing machine learning models. Data professionals and enterprise data teams need to know how this affects who is building models and what support they need. 

Familiarity with the machine learning lifecycle is needed because some stages of the machine learning cycle are easier to automate than others. Some parts of that lifecycle—such as defining the problem—cannot be automated and need human support. During the modeling phase, AutoML is highly successful at ranking potential solutions by model accuracy, but care must also be taken to evaluate models on business criteria. After deployment, AutoML can save time and effort in model monitoring and maintenance.

Accuracy is one crucial factor in a successful machine learning project, but having models that are difficult to interpret sometimes seems like trading one problem for another. AutoML can heighten this problem because it automatically chooses the most accurate models. The quest for accuracy often prompts data teams to consider complex “black box” models, yet analytics management and regulatory agency need transparency into the model’s prediction. In recent years, new solutions have been found to address this quandary. An explanatory layer can be built on top of black box models. Explainable AI, usually referred to as XAI, is a set of techniques for providing these explanations. 

In this one-day course, machine learning specialist Keith McCormick dives into what AutoML technology can and can’t do, as well as overview terms and concepts to understand XAI better. The focus will be practical traditional machine learning examples drawn from established industries. Participants will also see these concepts applied in some brief examples using a credit scoring dataset. There will be no software exercises, but there will be opportunities for workshop exercises and discussion. Data teams will leave with a better understanding of how to support this technology and knowing how to speak the language of AutoML when collaborating with data scientists. So, join veteran analytics and data science expert Keith McCormick for this one-day workshop.

Learning objectives

What you will learn:

• What is AutoML
• Navigating your analytics software options
• What AutoML can and can’t do at key stages in the lifecycle
• How AutoML chooses the “best” model, and what additional criteria should you apply in evaluating models
• What is Explainable AI (XAI), and why is it needed
• What are the most common XAI techniques, and how are they used
• What is the difference between Local and Global explanations, and when should you use them
• How to build the best possible transparent models to avoid the need for XAI.

Who Should Attend

• Business and IT professionals who want to understand AutoML better
• Analytics leadership who want to develop an appropriate strategy for both technology and team composition
• Data scientists and engineers will learn about the current state of the art and the recent rise of XAI
• Experienced DW/BI professionals be able to put AutoML and XAI techniques into a more complete enterprise data support context.