Data Science for Business: What You Need to Know about Data Mining and Data-Analytic Thinking

by Foster Provost

SVMs choose based on a simple, elegant idea: instead of thinking about separating with a line, first fit the fattest bar between the classes.