Having decided what to measure, identified suitable proxies, defined the Proxy Hypothesis, understood the accuracy and precision or the measurements, identified errors and – where possible – corrected them, it is possible to begin the analysis and simulation. Both require measurements: analysis is performed directly on the measurement data, whereas simulation uses the measurements for configuration and/or calibration.
The methods and examples illustrated below are to be used by an analyst in studying, understanding and extrapolating the measurements obtained. This specialized activity requires a high degree of numerical competency and familiarity with the subject area.
These methods contrast results incrementally often building on earlier results and ultimately always from the foundations of measurement. They do not in and of themselves provide insight into what to do, or which scenario to choose – although they can of course be used to support those kinds of decisions. However, this section is focussed on what can be inferred or deduced from the measurements taken independently of what ultimate purpose the analysis may be put to. For the application of the analysis to decisions making, see the next section on reporting and evidence.
There are many analyses that can be performed using the range of measurements that have been defined. The analytics that should be used depend on the measurements that are available and the kind of evidence required or decisions to be made. The following sections explore just a few of the more common and perhaps useful occupancy analyses, and provide some potential uses and considerations for use.
Most research into the use of buildings relies on data collection form real buildings in use. It is possible to collect large amounts of detailed data, describing real-world settings.
Simulation studies, on the other hand, are highly simplified and artificial. Simulation sacrifices vast amounts of real world data in exchange for simplicity, control and experimentation.
Simplicity means reducing the number of variables in a model, so that their interactions can be studied and compared and understood. Three or four interacting variables is plenty; a model with more than five or six variables is too complex to understand and therefore self-defeating.
Control is a consequence of simplicity – the modeller can set the values of all the variables and observe the outcomes. This means that the modeller can carry out experiments by changing attribute values in a systematic way, revealing trends and patterns in the outcomes.
ST – simulation of a modern organization
The ST (space-time) model is a simple but useful example of simulation modelling. It describes office-based organisations in terms of basic attributes that should be applicable in practically all situations. ST uses a simplified description compared to the incredible variety of the real world, but it has a specific purpose: understanding the use of space and time in a world where individuals have much greater freedom of choice about the times and places for carrying out activities. This decentralisation of decision-making brings unpredictability and complexity. Because individual decision-making is the critical factor, this is what the ST simulation model concentrates on.
The organisation to be modelled is described by the number of employees in nine employee types, defined by time and workstyle. In the time dimension, employees can be static, spending 90% of their working time at the employer’s premises; flexible, with 50% of their time on-site; or mobile with 20% of their time on-site (these percentages can be changed). The number of employees in each time category is input data for the ST model.
In workstyle, employees an be territorial, always using an assigned workspace that they ‘own’; or task-focused when they spend 75% of their time at the employer’s premises in bookable workspaces for individual work and 25% of their time in informal spaces for interaction; or interaction-focused when 25% of their time is in bookable workspaces and 75% in informal spaces (again, the percentages can be changed). The number of employees in each workspace category is also input data for the ST model. There are three workspace types – assigned (or allocated), shared (or bookable) and informal. The number of workspaces of the three types is the final element of input data for the ST model.