With the emergence of automated techniques for creating predictive models – sometimes called Predictive AI — such as autoML frameworks and Generative AI as leading-edge and game-changing technology, it’s natural for data scientists to ask the question, “Can these approaches help me?”. In this workshop, the question will be answered including both “pro” and the “con” sides of using autoML and Generative AI for predictive. Examples with well-known datasets will be used to illustrate the concepts, and all prompts and code used in the workshop will be made available to attendees.

This workshop will include live demos and participant interaction to improve models.

6 hours: 4 1.5 hour parts

Part 1: Definitions

1. Different Flavors of AI

• Predictive AI -> Predictive Analytics -> Data Science -> Machine Learning

1. Generative AI: Deep Learning and LLMs
2. Why predictive AI is only a new marketing label for pre-existing technology
3. Automation: AutoML
4. Example Use Cases for illustration

Part 2: AutoML

1. What goes into AutoML?
2. What isn’t in AutoML?
3. How does AutoML connect to Predictive AI and Generative AI?

Part 3: Build a predictive model using Generative AI (LIVE DEMOS)

1. Define the problem: Business Understanding
2. How Large Language Models can be applied for predictive use cases.
3. Hands-on exercise

• Prompt

1. Run code
2. Identify deficiencies
3. Repeat
4. Summarize results

Part 4: Comparing AutoML and GenAI to Human-driven Predictive AI (LIVE DEMOS)

This exercise will involve a different modeling software tool than in Part 3.

• Use prebuilt workflow elements but all “by hand”

1. Run model. Assess results
2. Identify deficiencies
3. Compare to Generative AI above

Conclusions:

1. What does AutoML and Generative AI for Predictive AI do well? What don’t they do well? Why or why not?
2. What are ways the best of AutoML and Generative AI can be used in Predictive AI?

Workshop Schedule:

• 09:00
  • Workshop program starts
• 10:30 – 11:00
  • Morning Coffee Break
• 12:30 – 13:30
  • Lunch
• 15:00 – 15:30
  • Afternoon Coffee Break
• 17:00
  • End of the Workshop

Intended Audience: Machine learning / predictive analytics practitioners who are interested in automating traditional approaches to building predictive models with new autoML and generative AI technology.

Knowledge Level: Prior experience with data preparation and/or building supervised or unsupervised learning models, either either using programming languages like R, Python, or SAS, or GUI-based tools like KNIME, RapidMiner, or WEKA. Participants are expected to have a baseline understanding of summary statistics and a few predictive modeling algorithms such as linear regression, logistic regression, decision trees, neural networks, and tree-based ensembles.

About the instructor

Dean Abbott is Chief Data Scientist at Appriss Retail and President of Abbott Analytics. Last year, he served as the Bodily Bicentennial Professor in Analytics at UVA Darden School of Business. He is an internationally recognized thought leader and innovator in data science and predictive analytics with more than three decades of experience solving a wide range of private and public sector problems. Mr. Abbott is the author of Applied Predictive Analytics (Wiley, 2014) and coauthor of The IBM SPSS Modeler Cookbook (Packt Publishing, 2013).