This is the dwelling dimensions dataset referred to in Özer, Seyithan, 'Interior Complex: Design Standardization in London's Housing', PhD diss., Royal College of Art, 2021. The dataset contains dimensional and spatial data derived from a sample of housing floor plans collected from inner London (n=2,007). The initial sample of 5,278 dwelling unit plans were converted into dimensional and numerical data using machine learning algorithms provided by two companies, Archilogic and Archilyse. The floor plans were first digitized with the machine learning-based algorithm Archilyse provided, and then the data was extracted from these models with the algorithm Archilyse provided.

For each of the 5,278 floor plans collected, Archilyse provided room-level data including 1) the net floor area, 2) the dimensions of the minimum bounding rectangle (width and length), 3) the circumference, 4) the total window length, 5) the number of doors (including the IDs of rooms the doors open on to) and 6) the number of kitchens, bathroom elements and staircases. A ‘room’ is defined as a space bounded and separated from others by walls and connected to others by doors. Therefore, rooms that are separated from each other by openings other than doors were counted as only one room. For instance, connected living and dining areas or entrance halls partially separated from living rooms were counted as single rooms. At the same time, built-in storage, which meets the criteria of a ‘room’ (enclosed by walls and separated by a door), were counted as separate rooms in the raw dataset. In its final form, a total of 1,840 plans were eliminated from the initial dataset of 5,278 floor plans. It is important to point that most of the eliminations were in houses and maisonettes of the older housing stock (Appendix A, see Related Sources and documentation available in the bundle).