Why Contractors Should Invest in a Vertical Analytics Platform

A vertically-focused data analytics platform is a no-brainer for today’s modern construction enterprise.

Narrow Domains

The world of machine learning is vast and exciting and runs the gamut from extremely valuable use cases to more fringe use cases. For example, IBM’s Watson AI was fed 15 years of award-winning car commercials and asked to create one of its own. Is this the groundbreaking science the world needs? Perhaps not.

Luckily, machine learning is emerging as a tool of real value in numerous businesses. Many of the Fortune 500 firms employ machine learning to speed up and optimize tasks in their day-to-day business activities. From UPS and FedEx using software to optimize loading and packing pallets, to clothing retailers using RFID sensors on the ceilings of their stores to ensure and track that every piece of merchandise stays in the correct location.

Using data analytics and visualizations to influence better decisions is the future of the modern enterprise– but not just any data tool will work. Are you a general contractor trying to do analytics on a horizontal platform like Power BI or Tableau? If so, you may be hindering your data strategy.

When it comes to the future of AI-powered data analytics, the ability to narrow the domain of the machine will return superior results to those of a general domain. In other words, the difference between a vertically focused machine learning platform and a horizontal machine learning platform can make or break your data strategy.

Learn by Example

There is a dizzying amount of data being created in your company and it is nearly impossible for any human to read, analyze and make sense of all of it. No human has time to watch 15 years’ worth of car commercials, nor does a human have the stamina to monitor the packing and loading of every pallet in a FedEx fleet.

Faster, cheaper storage gives us the ability to capture and store this data, while machine learning gives a computer the ability to analyze, evaluate and anticipate outcomes using that data. Machine Learning essentially allows computers to do things intelligently and at scale by relying on a framework of statistical models that perform actions based on pattern recognition rather than human intervention. These algorithms are manifested in a number of ways, but they all “learn” how to perform some action, whether it is reading documents, automating workflows, or creating predictive analytics on data sets.

The “teaching” process is where a lot of companies will fail in getting good results. There are many out of the box solutions that promise machine learning with data, but many of these platforms are horizontal providers. In other words, they serve many industries; companies from logistics, healthcare, and media may all be having access to these algorithms in the hopes of seeing predictions in the data. Unfortunately, their chances of seeing anything of value are limited.

“It’s very difficult to translate learnings across domains,” says John Cowgill of Costanoa Ventures. “Neural nets work best in a highly confined domain where they are trained to recognize a specific set of signal data in a massive, noisy data set. They often struggle mightily in more general problems.”

A machine that learns how to direct a car commercial would be disastrously bad at detecting anomalies in a Request for Information (RFI) written by a subcontractor. If data types have nuances, so too should the data platforms that store and analyze that data.

Headlines like “Vertical Beats Horizontal in Machine Learning” and “Horizontal Strategy is Out” echo this trend. William Vorhies, the long-time Editorial Director at Data Science Central, believes the benefits of horizontal machine learning is fading. “Horizontal companies don’t own the customer’s core problem. They simply provide a tool that must either be adapted by consultants familiar with that industry or requires the client customer to learn more about [data science] than they probably wish to.”

A company should choose their data analytics platform based on the data sets and focus of the provider, not necessarily by brand name. When it comes to data science in today’s world, vertical is truly better.

Ellis Talton