A four-hour, hands-on workshop at KCDC 2022 that takes a room with no background from the vocabulary of machine learning to building models themselves, entirely in Google Colab so nobody fights an install.
It runs the full arc: the terms and problem types (and the traps, like overfitting and data leakage), data wrangling in pandas and NumPy, regression and then classification in scikit-learn with the metrics that tell you whether a model is any good, and a close on neural networks in TensorFlow and when they earn their keep. The datasets are real ones — car sales, heart disease — rather than toy arrays.
The materials are on GitHub.