Space planning is an important aspect of architectural design. It is the process by which available space is differentiated into zones for particular purposes. In existing literature, a wide range of computational methods have been applied to space planning with a view to automating the process. However, such research was limited by tractability and knowledge extraction issues, and a lack of research into how designers use automated design tools in practice. To address the tractability and knowledge extraction issues, a general data-driven direction was adopted for this research. In this thesis, contemporary residential flat layout design in the UK was studied as an example problem in space planning. The problem was addressed in two separate tasks. The first task was estimating the feasibility of flat types to obtain a realistic brief for a given footprint. The second task was to generate a layout based on a brief for a given footprint. A prototype program was built for each of these two tasks. Prototype Program 1 used a Random Forest classifier that learned from a dataset and estimated flat types for a given footprint. Prototype Program 2 applied a novel case-based learning framework. It used the shape and context of footprints as an index to retrieve existing cases as grating layouts and then used an optimisation method to create new layouts for the given footprints. To research how designers interact with automated design tools, the final part of this thesis reports the design and results of experiments conducted with a panel of eight designers. The experiment examined human-computer interactions with the designers under two contrasting models using a Wizard-of-Oz technique. The designers’ responses were recorded in interviews and analysed in detail. This research contributes to the understanding of data-driven space planning, particularly to the expanded use of grating representation and footprint shape-context as an indexing method. The study of the interaction between human designers and computer software provided insight into vi important issues such as design agency, explainability and design exploration, when integrating automated tools into design practice.