Applied machine learning
Undergraduate · Data science
Syllabus focus
Topics typically covered
Standard syllabus
Supervised learning core
- Problem framing: prediction vs explanation
- Linear models and regularization (ridge/lasso intro)
- Trees, random forests, and boosting survey
- Distance-based and naive Bayes survey
- Feature preprocessing: scaling, encoding, imputation
- Cross-validation and hyperparameter search (intro)
Evaluation and diagnosis
- Classification and regression metrics
- Class imbalance strategies
- Learning curves and under/overfitting diagnosis
- Calibration and probability outputs (intro)
- Error analysis by slice / subgroup
- Baseline models and sanity checks
Workflow and MLOps intro
- Train/validation/test discipline and leakage
- Pipelines and feature stores (survey)
- Model serialization and versioning
- Monitoring drift concepts (intro)
- Experiment tracking survey
- Documentation: model cards (intro)
STEM / applied
Implementation
- scikit-learn or equivalent course stack labs
- End-to-end tabular competition-style project
- Text or image pipeline survey (optional)
- Neural nets at awareness level unless course-required
- Interpreting SHAP/feature importance carefully
- Cost-sensitive decisions and threshold tuning
Responsibility
- Bias, fairness metrics survey, and harms
- Privacy and membership risks (intro)
- Security of model artifacts
- Stakeholder communication of uncertainty
- When not to deploy a model
- Capstone: trained model + evaluation write-up
Notes
Shallower than graduate ML; depth follows the syllabus (classical ML vs light deep learning). Overlaps CS ML intro when both exist.