Summary
In this chapter, we introduced ML models as problems of mathematical optimization or mathematical programming. We found out that an end-to-end ML project is the sum of multiple small optimization problems. We also gained knowledge about how businesses can unlock the true value of data upon leveraging mathematical models (primarily driven by mathematical equations) in addition to ML (driven by data) models. We learned that an ML model is predominantly a predictive tool and a mathematical model is a prescriptive one.
In the next chapter (which begins the next part of the book), we will take a meticulous look at a well-known algorithm called PCA, utilized in an unsupervised ML model fit to data with high dimensionality. It is a dimensionality reduction technique and one of the most tried and tested mathematical tools employing constrained optimization.