Learn From Data

Linear algebra, calculus, probability, statistics, entropy and optimization

• Vectors, matrices, PCA & SVD
• Derivatives & gradient descent
• Bayes, covariance & correlation
• Entropy & hypothesis testing

The four families of ML, choosing algorithms and metrics, Deep Learning, GenAI & LLMs

• The four families of ML
• Metrics & cross-validation
• Deep Learning (CNN/RNN)
• Generative AI, LLMs & Agents

MSE, the Normal Equation, gradient descent, assumptions, residuals and R²

• The model & the MSE loss
• Normal Equation & gradient descent
• Four assumptions & residuals
• R², RMSE and polynomial regression

Sigmoid, binary cross-entropy, Softmax and the classification metrics

• Sigmoid & decision threshold
• Binary cross-entropy
• Softmax & One-vs-Rest
• ROC-AUC & class imbalance

Missing data, scaling, bias-variance, Ridge/Lasso/Elastic Net

• Three missingness mechanisms
• Scaling & encoding
• The bias-variance trade-off
• Filter / Wrapper / Embedded

Gini/Entropy, CART, controlling overfitting, CV and hyperparameter tuning

• Gini, Entropy & Information Gain
• The CART algorithm
• K-Fold & Stratified CV
• GridSearch & RandomizedSearch

Bagging, Random Forest, AdaBoost, Gradient Boosting & XGBoost

• Bootstrap, Bagging & OOB
• Random Forest
• AdaBoost & Gradient Boosting
• XGBoost

XOR, MLP, forward pass, backpropagation & vanishing gradients

• XOR & neuron = Logistic Reg
• MLP, forward pass & activation
• The 4 backpropagation equations
• Vanishing gradients & initialization

Lab: NumPy from scratch and Keras (TensorFlow)

• NumPy scratch — manual backprop
• Three ways to build a Keras model
• model.fit() & GradientTape
• TensorBoard