Imagine you’re teaching a friend to recognize a handwritten digit. One approach: sit down and invent rules like “a 7 has a horizontal top bar and a slanted stem.” Another approach: just keep a pile of example digits. When a new drawing arrives, compare it to every example and pick the closest match. The first strategy builds a compact model from the data; the second simply remembers the data and acts at the last moment. In machine learning, algorithms that follow the second path are called lazy learners — and k‑nearest neighbors is the classic one. Support Vector Machines (SVM), on the other hand, build a powerful geometric model that separates classes with a clear margin. This chapter explores both families, showing you when a simple nearest‑neighbor lookup works well and when drawing a clear boundary is the better choice.