Machine Learning in Python, Inspired by Andrew Ng

Note: the course is originally taught in Oracle/MATLAB. Implementations in Python are made by me. Data sets are taken from Andrew Ng's course

Exercise 1 : Linear Regression

Univariate LR

• Data Visualizations
• Coding Cost functions and gradient descent from scratch
• Visualizing J($\theta$) cost

Multivariate LR

• Data Visualizations
• Feature Normalization
• Make predictions

Exercise 2 : Logistic Regression

• Data Visualizations
• Cost functions and Gradient code from scratch
• Feature normalization
• Apply logistic regression
• Apply regularization
• Visualize boundary of the classification

Exercise 3 : Multi-Class Classification

• logistic regression Cost and Gradients (multi-class)

One vs All

• OVA and noBias OVA code without built-in OVA functions
• Make predictions

Neural Networks

• Code a neural net predictor
• Make predictions

Exercise 4 : Neural Networks Learning

• Coding Neural net Cost functions from scratch
• Initialize weights (randomly)
• Code gradient descent for Neural nets
• Make predictions using said neural net

Exercise 5 : Regularized Linear Regression and Bias v.s. Variance

• LR Regularized Cost and Grad
• Learning curve visualizations
• map data into polynomial feature space
• Visualize different polynomial fits
• Get learning curves for Training and Cross-Validation