A major task of every oil company is oil field optimization, i.e. maximizing oil production and reducing operational cost. Knowledge about injector-producer relationships (IPRs) is crucial for optimal operation of oil fields. However, inferring IPRs has been a challenging problem due to the unknown underlying structure of oil fields, continuous change of the underlying structure over time, and the large number of wells, i.e. typically, hundreds of injection wells and hundreds of production wells. This book provides two different approaches which map the IPRs problem to a large-scale parameter estimation problem. One approach is constrained nonlinear optimization and the other is machine learning approach. The two approaches demonstrate that not only prediction accuracy but also computational efficiency can be achieved for large-scale parameter estimation problems. This book should help field engineers optimally operate oil fields and show researchers practical examples about how to apply optimization and machine learning techniques to oil field optimization.