Machine Learning by Andrew Ng (Coursera)

The original course used MATLAB or Octave and became one of Coursera's landmark early offerings. It was later succeeded by the three-course Machine Learning Specialization, which teaches exercises in Python.

The course begins with linear regression and logistic regression, then develops topics such as regularization and supervised classification. This progression establishes cost functions and gradient descent before the neural-network material.

Basic algebra and comfort reading equations are enough for much of the introductory material, because Ng explains the optimization ideas as they appear. Some familiarity with derivatives helps when studying gradient descent, but the course is not built as a proof-heavy mathematics class.

The Coursera course is designed as an accessible introduction with guided programming work. Stanford's CS229 is a more mathematically demanding university course that expects stronger probability, calculus, and linear algebra.