Browse machine learning topics
Move from data and evaluation foundations into supervised learning, unsupervised learning, model selection, and practical scikit-learn workflows.
Topics
Available topics
Move from data and evaluation foundations into supervised learning, unsupervised learning, model selection, and practical scikit-learn workflows.
Machine Learning Foundations
Build the core mental models for what machine learning is, how models learn from data, and how classical machine learning fits within the wider AI landscape.
Go learning pathData Preparation and Pipelines
Prepare tabular data for reliable modeling by defining features and targets, handling data types and scale, preventing leakage, and building repeatable preprocessing workflows.
Go learning pathFeature Engineering and Representation
Understand how feature choices, transformations, geometry, sparsity, and dimensionality shape what classical models can learn and how reliably they can generalize.
Go learning pathRegression and Classification Models
Understand the main classical supervised-learning model families through their objectives, assumptions, inductive biases, and practical boundaries.
Go learning pathTree Models and Ensembles
Learn how decision trees partition data and how bagging, randomization, and boosting combine trees to change variance, bias, robustness, and interpretability.
Go learning pathClustering and Dimensionality Reduction
Investigate unlabeled structure with clustering and dimensionality reduction while making geometry, scaling, validation, and interpretation limits explicit.
Go learning pathValidation and Model Selection
Compare models fairly with honest data boundaries, baselines, cross-validation, tuning, reproducible experiments, and a disciplined selection workflow.
Go learning pathEvaluation and Model Diagnostics
Measure predictive quality and diagnose failure with metrics, error analysis, class-imbalance reasoning, generalization diagnostics, thresholds, calibration, and interpretation.
Go learning pathModel Persistence and Inference
Move fitted classical models from experiments into repeatable prediction workflows: preserve preprocessing, save and load artifacts, maintain input-schema and version compatibility, and verify inference behavior without expanding into full MLOps or deployment infrastructure.
Go learning path