A Machine Learning Cookbook Where Some Recipes Are Still Blank

What it does

This repository is a curated index of machine learning tutorials built from scratch in Python. It organizes implementations and project ideas across nine categories—core algorithms, neural networks, computer vision, NLP, recommendation systems, time series, anomaly detection, sentiment analysis, and miscellaneous applications—ranging from linear regression to transformer-based LLMs. Some entries link to concrete Python files, while many others remain listed as future tutorials without attached code.

The interesting bit

The collection treats “no black boxes” as a full curriculum: you are meant to write gradient descent, backpropagation, and even a “mini TensorFlow” using raw NumPy rather than calling library methods. The scope is deliberately maximalist, stretching from K-Means clustering to diffusion models and GANs.

Key highlights

• Lists dozens of topics across nine categories, from linear regression to GANs and LLMs
• Some tutorials include linked Python implementations (e.g., linear_regression.py, activation functions, KNN)
• Also catalogs applied projects: recommendation systems, OCR, sentiment analysis, stock prediction
• Maintained by Outcome School as a free, project-based learning resource
• Explicitly noted as an ongoing work in progress

Caveats

• Many listed tutorials are placeholders without linked code or implementations yet
• The README is mostly a table of contents; depth and quality of existing scripts are unclear without inspecting individual files
• Heavy self-promotion of Outcome School courses throughout the page

Verdict

Worth bookmarking if you learn best by re-implementing algorithms in Python and NumPy. Skip it if you need a fully finished course or a unified framework rather than a grab bag of discrete tutorials.