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 detection in online social networks and weather prediction. For spam detection, we discuss the limitations of existing ML-based methods that are implemented in real-world scenarios and propose an effective spam detection solution equally applicable to all social networks, relying on network-specific attributes and available APIs, for superior spam-capturing performance. For weather prediction, We introduce the Micro-Macro Model (MiMa), a model tailored for short-term, location-specific weather parameter prediction. The proposed approaches can be extended to all other real-world problems that can follow this framework by treating problem-specific data handling as their key components. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://bibliotheek.ehb.be:2222/en-US/products/dissertations/individuals.shtml.]