BIMcrunch has become a platform for a plethora of elite names from the Building Information Modelling sector to become guest writers on the site and share their thoughts and opinions with the entire #GlobalBIMCrew.
In the first BIM Voice instalment of 2016, it is our pleasure to welcome Brett Young, CEO and Founder of BIM system coordination company, BuildingSP. Brett has more than 15 years of experience in the BIM field and is currently utilising generative design as part of his business’ BIM processes.
In this opinion piece, Brett looks at how data mining for BIM can be compared to… buying diapers and beer?! You will see how his analogy is absolutely accurate by reading below.
The AEC industry is at a critical juncture in history. BIM, cloud computing, and the growing sophistication of our building products have created the perfect environment for a new era in AEC history. Autodesk calls this “The Era of Connection,” a disruptive time when design, construction, and operational processes will be radically changed. This is an exciting time. There is a parable from the business intelligence industry that is highly relevant to AEC in 2016. This parable is about two very familiar consumer product types: diapers and beer.
The odds are better than average that you have a firm understanding, if not an emotional connection, to either diapers or beer (or both). The odds are very low that you have a similar connection to nascent concepts like “generative design.” The purpose of telling this parable is, therefore, not to point out the significant contributions to mankind that either diapers or beer have provided and compare them to growing AEC trends. The purpose is related to connection. Here goes:
In the early 1990s, a grocery store chain in the Midwest accumulated 1.2 million transaction records of customers checking out at their stores. Each of these records represented a “basket” of purchases, or the total of one customer’s purchases made at the cashier. Imagine 1.2 million shopping carts, filled with groceries.
The accumulation of these transaction records was made possible by advances in point of sale systems, bar codes, and databases. Given these new tech advances, what is a grocery store chain to do with all this data?
Hire a consultant.
Enter Teradata, managed by Thom Blischok. In database operations, a “query” is a question that you ask of a set of data. Thom’s team had a different approach. They didn’t ask questions of the data. Instead, they looked for correlations in the data. They agnostically looked for periods when high-profit items were correlated into “baskets.” The key distinction here is that they didn’t ask questions – they correlated the unknown.
What did they find?
Diapers and beer. On Friday nights, between 5:00 p.m. and 7:00 p.m., profits spiked when the two product types were purchased together. Diapers were positively correlated to beer sales. The consumer behavior driving this phenomenon is related to the end users of diapers. If you are buying diapers, you have a kid at home. If you have a kid at home, you’re probably not able to go to a bar and drink beer. You are driven to a consumer choice – buy some beer – to replace this behavior. This is all succinctly summarized by the grand understatement: “Offspring acquisition is a known carousing inhibitor.”
Given this new insight, the grocery store chain could have placed the beer closer to the diapers to encourage inhibited carousers. This technique did not occur at this grocery store chain, but absolutely continues to occur with other product types.
The parable of diapers and beer is used as an easily understood example of “data mining,” where correlations are explored rather than data queried. It’s a parable because some of the facts are a little fuzzy and it’s meant to be instructive.
How is this related to AEC?
“Generative design” is the use of computational tools to create designs and design models. At BuildingSP, we use generative design to create MEP layouts programmatically in 3D, without clashes, and in ways that mimic how we manually route systems. We generate lots of solutions for building infrastructure layouts by applying machine learning to the parameters that drive our algorithms. These sets of solutions can then be compared, scored, and evaluated.
Generative design creates the opportunity to apply data mining to BIM, both to new sets of designs and existing models. We’ve already gained insight from the application of generative design to MEP systems. But we hope to soon find our “diapers and beer” moment and make correlations, enabled by technology.
Visit BuildingSP’s official website to learn more about their services.