Exploring Optimization Techniques

This chapter primarily aims to address the question, “Why is optimization necessary while solving problems?” Mathematical optimization, or mathematical programming, is a powerful decision-making tool that has been talked about in depth in the chapters of Part I. What is important is to recall the simple fact that optimization yields the best result to a problem by reducing errors that are, essentially, the gaps between predicted and real data. Optimization comes at a cost; almost all optimization problems are described in terms of costs such as money, time, and resources. This cost function is the error function. If a business problem has clear goals and constraints, such as in the airline and logistics industries, mathematical optimization is applied for efficient decision-making.

In machine learning (ML) problems, the cost is often called the loss function. ML models make predictions about trends or classify data wherein training...