It is the ability to efficiently search through the design space of any given problem (i.e. real project) that is the holy grail of what we nowadays call parametric optimization.
assessing the whole design space is practically, and not mathematically, intractable.
The traditional solution to this is linked to our ability to optimize mentioned above. People would use their past experience to prune off certain parameters and/or values. Ultimately what this does is reduce the design space to a handful of alternatives that can be then assessed. There is one problem with traditional approaches, like most traditions they cling strong to the past and propagate standard practices.
is there a way to populate the vast design spaces of building models in a feasible and practical way?
‘machine learning is like the new electricity‘
What would Machine Learning for building design look like?
AEC talent, according to the chart, is hard to come by – not a lot of data scientists working in the architecture, engineering and construction space
urban planners can benefit significantly from the analysis of personal location data.
We have less data than other sectors
We have fewer stakeholders than other sectors, and
We are also less transaction-intensive.
Productivity in the AEC industry ought to increase in the near term, when new technologies such as building information modeling (BIM), collaborative work processes such as integrated project delivery (IPD), lean construction, and now – data – are used seamlessly and comprehensively throughout design, construction and building lifecycle.
As architecture firms drive out those interested in connecting design, construction and client driven data to services, clients are doing the opposite – especially in retail where big data is driving processes from capital planning to store rollout/renovation and management.