Machine Learning

86 topics across 7 chapters

Chapter 1

Foundations of Machine Learning

ML workflow & terminology

Math essentials for ML

3 subtopics

Linear algebra drills (vectors, matrices, dot products)

Calculus & gradients (partials, chain rule intuition)

Probability basics (distributions, expectation, Bayes rule)

Programming for ML (Python tooling)

3 subtopics

Python + NumPy basics (arrays, broadcasting)

pandas data wrangling (joins, groupby, time series basics)

Visualization for ML (matplotlib/seaborn, diagnostic plots)

Problem framing (objectives, constraints, baselines)

Chapter 2

Data & Feature Engineering

Data collection & labeling (quality, sampling, guidelines)

Cleaning & preprocessing

3 subtopics

Handle missing values & outliers (imputation, robust stats)

Categorical encoding (one-hot, target encoding pitfalls)

Scaling & normalization (standardize, min-max, when/why)

Feature engineering

3 subtopics

Text features (bag-of-words, TF-IDF)

Feature selection (filters, wrappers, embedded methods)

Leakage prevention checklist (pipelines, target leakage patterns)

Train/validation/test splits (including time-aware splits)

Chapter 3

Supervised Learning

Linear models

2 subtopics

Linear regression practice (least squares, assumptions, diagnostics)

Logistic regression practice (odds, decision boundary, calibration)

Tree-based models

3 subtopics

Decision trees (splits, depth, pruning intuition)

Random forests (bagging, OOB error, feature importance caveats)

Gradient boosting concepts (GBDT intuition, overfitting controls)

Support Vector Machines (SVMs)

k-NN & instance-based learning

Regularization & bias-variance

3 subtopics

L1/L2/Elastic Net exercises (effect on weights, sparsity)

Early stopping intuition (learning curves, patience, checkpoints)

Bias-variance tradeoff examples (underfit vs overfit case studies)

Imbalanced learning (resampling, class weights, thresholding)

Chapter 4

Unsupervised & Representation Learning

Clustering

3 subtopics

k-means (objective, initialization, choosing k)

Hierarchical clustering (linkage, dendrogram reading)

DBSCAN (density, epsilon/minPts intuition, failure modes)

Dimensionality reduction

3 subtopics

PCA (variance explained, reconstruction error intuition)

t-SNE/UMAP intuition (what plots do/don't mean)

Matrix factorization basics (SVD intuition, embeddings)

Anomaly detection basics (distance/density methods, evaluation pitfalls)

Recommender systems

3 subtopics

Collaborative filtering (user-user/item-item, cold start issues)

↗ Matrix factorization basics (SVD intuition, embeddings) (see Chapter 4)

Ranking metrics basics (MAP, NDCG, offline vs online mismatch)

Chapter 5

Deep Learning

Neural network fundamentals

3 subtopics

Backpropagation intuition (computational graphs, gradients)

Activations & initialization (ReLU family, saturation, variance)

Common loss functions (MSE, cross-entropy, contrastive basics)

Convolutional Neural Networks (CNNs)

Sequence models & Transformers (overview)

Training & optimization tricks

3 subtopics

Optimization with SGD vs Adam (learning rate schedules)

Batch norm & residual connections (why they help, intuitions)

Hyperparameter tuning basics (search spaces, overfitting to val)

↗ Regularization & bias-variance (see Chapter 3)

Generative models overview (autoencoders, diffusion, GAN intuition)

Chapter 6

Model Evaluation, Deployment & Responsible AI

Metrics & error analysis

3 subtopics

Classification metrics (precision/recall, ROC-AUC/PR-AUC, thresholds)

Regression metrics (MAE/RMSE/R², residual analysis)

Calibration & uncertainty basics (reliability curves, confidence)

Cross-validation & experiment tracking (principles and practice)

Model serving basics

3 subtopics

Batch vs real-time inference patterns (latency, throughput tradeoffs)

Packaging models (Docker + artifact formats like ONNX basics)

Inference APIs (REST/gRPC basics, input validation, idempotency)

Monitoring & drift detection (data/label drift, alerting)

Reproducibility & versioning (data/model/code lineage)

Responsible AI & ethics

3 subtopics

Fairness & bias (definitions, measurement, mitigations overview)

Privacy & security basics (PII, membership inference intuition)

Interpretability (SHAP/LIME intuition, global vs local explanations)

Chapter 7

Applied ML Projects

End-to-end tabular ML project (EDA → model → report)

Text classification project (spam/sentiment with a strong baseline)

Image classification project (transfer learning baseline)

Time-series forecasting project (leakage-aware evaluation)

Kaggle workflow basics (feature pipelines, CV, ensembling hygiene)

Communicating ML results (model cards, limitations, stakeholder docs)