Machine learning with Python

Jason Brownlee's Step-By-Step Machine Learning In Python, from Machine Learning Mastery, is the standard first project: load the iris dataset, explore it, and compare several scikit-learn models. Solving A Fruits Classification Problem and Predict Quality Of Wine follow the same workflow on different datasets and add feature scaling and model evaluation.

To understand what the libraries do, write the algorithms yourself. Write Linear Regression From Scratch in Python is a video, and Linear Regression in Python without sklearn and Multivariate Linear Regression without sklearn implement gradient descent with NumPy. Build Your Own Neural Net from Scratch in Python goes one step further with a small two-layer network and backpropagation.

Applied projects round out the section: a Music Recommender using KNN, Find Similar Quora Questions (2 parts) with bag-of-words, TF-IDF and a random forest, and Detecting Fake News with Python and Machine Learning, a text classification project. Learn Unsupervised Learning with Python points to scikit-learn's own guide to clustering and dimensionality reduction, which is reference material rather than a project.