machine_learning_basics

Why we included this project

This is a study resource rather than a toolbox. Each notebook reimplements a classic algorithm from scratch in plain Python, with no scikit-learn or other machine learning libraries doing the work, so every prediction step stays readable and traceable. Coverage spans linear and logistic regression, decision trees, k-nearest-neighbor, k-means, support vector machines, PCA, perceptrons, and a simple neural network, alongside notebooks on preprocessing numerical/categorical data and images. You can run the notebooks in the browser through Binder without installing anything, though the preprocessing ones require downloading a dataset first. The author deliberately favors clarity over speed, which makes this a good fit for self-learners and for anyone walking a junior developer through ML fundamentals.