Machine Learning

6 subjects · 14 chapters · 56 topics

1. 01

   Foundations

   1.1 What machine learning is

   • Supervised, unsupervised and reinforcement
   • ML versus traditional programming
   • The ML project lifecycle
   • Where ML fails

   1.2 Maths for ML

   • Linear algebra essentials
   • Probability and distributions
   • Calculus and gradients
   • Optimisation intuition

   1.3 Tooling

   • NumPy, Pandas and scikit-learn
   • Notebooks and experiment tracking
   • Reproducible pipelines
   • Version control for data
2. 02

   Data Preparation

   2.1 Preparing features

   • Handling missing values
   • Encoding and scaling
   • Feature engineering
   • Train/test splitting and leakage

   2.2 Imbalanced and messy data

   • Resampling techniques
   • Class weights
   • Outlier handling
   • Data augmentation
3. 03

   Supervised Learning

   3.1 Regression

   • Linear and polynomial regression
   • Regularisation: ridge and lasso
   • Assumptions and diagnostics
   • Interpreting coefficients

   3.2 Classification

   • Logistic regression
   • K-nearest neighbours
   • Naive Bayes
   • Support vector machines

   3.3 Ensembles

   • Decision trees
   • Random forests
   • Gradient boosting: XGBoost and LightGBM
   • Stacking and blending
4. 04

   Evaluation and Tuning

   4.1 Measuring a model

   • Confusion matrix and derived metrics
   • ROC-AUC and PR-AUC
   • Regression metrics
   • Baseline comparison

   4.2 Improving a model

   • Cross-validation strategies
   • Grid, random and Bayesian search
   • Bias-variance trade-off
   • Learning curves
5. 05

   Unsupervised and Deep Learning

   5.1 Unsupervised methods

   • K-means and DBSCAN
   • Hierarchical clustering
   • PCA and t-SNE
   • Anomaly detection

   5.2 Neural networks

   • Perceptrons and activation functions
   • Training with backpropagation
   • CNNs for images
   • Transfer learning
6. 06

   Deployment and Project

   6.1 Putting a model into production

   • Saving and loading models
   • Serving predictions via an API
   • Monitoring and drift
   • Retraining strategy

   6.2 Capstone

   • Framing the problem
   • Building and evaluating
   • Deploying a demo
   • Reporting the results