The Exam in the Bin: Supervised Learning, Loss, Bias-Variance and Leakage

Deep End Era. Engineering with Sankalp · Oct 28, 2026 · 51m

Two months into his first job, Tomasz builds a hospital deterioration model with an area under the curve of 0.991. His manager Priya knows that the first question about a model that good is what it cheated with. Part one of The Deep End's Machine Learning series builds supervised learning from the ground up: the student, the past papers and the exam; the model family, the loss and the search; why squared error is an assumption about noise; the three students who draw a line through ten dots; bias and variance as an archer's grouping; double descent; and regularisation as a tailor's restraint…