Machine learning (ML) techniques have been successfully applied to a wide array of applications. This dissertation aims to take application data handling into account when developing ML-based solutions for real-world problems through a holistic framework. To demonstrate the generality of our framework, we consider two real-world applications: spam…