IBM IBSC 2026-04-08 | Module 7 Live Session | ML Regression, Classification & AI Engineering

In this live session for Module 7, we introduce the foundations of machine learning and how predictive models are built, evaluated, and eventually used in real-world data science workflows.

This week marks the transition from analyzing and visualizing data to using data to make predictions. We focus on supervised machine learning, including both regression and classification, and connect those ideas to the broader machine learning lifecycle: defining the problem, preparing data, training models, evaluating results, and thinking about how models are deployed into production.

We also spend time discussing the differences between data scientists, machine learning engineers, and AI engineers, including how ML models, batch inference, online inference, dashboards, RAG applications, agents, observability, and trust all fit into modern data and AI systems.

In this session, we cover:

• What machine learning is and how it uses historical data to make predictions
• Supervised vs unsupervised machine learning
• Regression vs classification and when to use each
• The machine learning lifecycle and how it connects to CRISP-DM and MLOps
• The difference between data scientists, ML engineers, and AI engineers
• How models move from notebooks into production systems
• Batch inference, online inference, dashboards, and downstream decision-making
• AI engineering concepts like RAG, agents, tools, guardrails, observability, and LLM gateways
• Risks and limitations of machine learning and AI systems
• Bias, trust, explainability, and why model outputs need to be evaluated carefully
• Using scikit-learn as a core machine learning framework in Python
• Linear regression for continuous numeric predictions
• Train/test splits and overfitting
• Model evaluation for regression
• Multiple linear regression and polynomial regression
• Why nonlinear problems often lead to tree-based models in practice
• Logistic regression as a classification model
• Decision boundaries and probability-based classification
• Classification evaluation concepts such as false positives, false negatives, and log loss

The big idea:

Machine learning is not just about picking an algorithm. It is about building a reliable workflow for turning data into predictions, evaluating those predictions, and using them responsibly in real-world systems.

Strong machine learning systems depend on good data, thoughtful feature selection, proper evaluation, and trust. Whether the final output is a dashboard, a production model, or an AI agent, the goal is the same: create useful, reliable, and explainable tools that help people make better decisions.

This week lays the foundation for more advanced machine learning topics in the following modules, including tree-based algorithms, random forests, unsupervised learning, model tuning, production machine learning, and advanced AI topics.

All notebooks and live session materials for this bootcamp are available here:
https://github.com/ABoothInTheWild/ibm_data_science

#datascience #machinelearning #python #regression #classification #scikitlearn #mlops #aiengineering #genai #analytics #bootcamp #ibm

Видео IBM IBSC 2026-04-08 | Module 7 Live Session | ML Regression, Classification & AI Engineering канала Alexander Booth