Data Mining Application for Architecture and Engineering Design

An unusual place to find innovative architecture, engineering, and construction industry news, Techcrunch.com introduced me toBerkeley, CA based Vernox Labs, a Y-Combinator backed company which recently released an interesting data mining product for the industry. The tool shows great promise to uncover, predict, and streamline design and construction issues for complex projects. I have no connection to the company but have great interest in all technologies which help me build.

Vernox Labs product is, essentially, a private searchable AEC industry database. One can see parallels here with the recent emergence of private medical databases for doctors. The article reports the company has spent resources collecting and cataloging all sorts of reports and emails from the AEC industry and thereafter analyzes it for trends. Perhaps the value of such a database to the AEC industry is best shown with an example: While a project is in the design phase one could query the database about a specific product or material. The system would then return information about the product's performance and installation, including any issues which might cause delays.

The application of big data and data mining to the AEC industry is complex, probably worth its own seminar or university degree. But needless to say I can already identify a couple variables which could affect the usefulness of Vernox Labs' tool which are unknown to me as of this writing. For one, the size of the database is of paramount importance. A large comprehensive database could actually be quite useful. But a skimpy database would be next to useless, returning only shallow results for any one query. Traditionally the AEC industry has been very protective of their internal communications and data because it can have such a large impact on construction pricing, fee structures, real estate and future projects so it will be interesting to see where Vernox Labs got their data from. The other dominating factor is the analysis and searchability of the data. Well structured, one could always hope to find what they are looking for and uncover trends which are not readily available without using mathematics. On the other hand, unstructured data might as well be a spreadsheet full of random numbers for how useful it is.

A final overriding factor, which dips into the realm of subjectivity, is one's ability to ask pertinent questions of the data. Good data analysis and getting good results from the data really is more art than science and if one can look at their data and ask really creative questions of it sometimes very valuable conclusions come tumbling out. The AEC industry has long been waiting for the application of big data to construction and Vernox Lab's seems to be worth watching for further details.