AINS6002: Machine Learning & Predictive Modeling

AINS6002: Machine Learning & Predictive Modeling#

Aurnova MSAI track: Core
Credits: 3
Format: 8-week online graduate course

Covers supervised, unsupervised, and operational predictive modeling with reproducible evaluation.

This course follows the Aurnova/Castalia course-site pattern used by AINS6003: each module includes book prose, an assignment notebook, slide notebook, narration, instructor notes, and an executable lab.

Course Outcomes#

By the end of the course, students will be able to:

  • explain the major concepts and tradeoffs in Machine Learning & Predictive Modeling;

  • build or evaluate applied AI artifacts aligned with the course domain;

  • document assumptions, evidence, limitations, and operational risks;

  • connect technical work to governance, stakeholder needs, and deployment readiness.

Module Map#

  1. Prediction tasks and data framing — What decision is the model actually supporting?

  2. Data preparation and feature pipelines — How do preprocessing choices shape model behavior?

  3. Linear and tree-based baselines — Why do strong baselines matter before complex models?

  4. Model selection and validation — How do we choose models without overfitting our evidence?

  5. Unsupervised learning and structure discovery — How can models reveal patterns without labels?

  6. Time, drift, and monitoring — What changes after a model leaves the notebook?

  7. Interpretability and stakeholder explanation — What explanations are appropriate for different audiences?

  8. Predictive modeling portfolio — What evidence supports a production recommendation?