IBM Watson Tackles Water Data with Machine Learning

Logistics
Machine Learning Detects Fuel Waste and Fraud
July 14, 2016
petri dish healthcare research
AI Reveals Bat Carriers of Ebola and Other Filoviruses
July 15, 2016

IBM Watson Tackles Water Data with Machine Learning

ibm microchip board

Some water utilities have so much data at their disposal it is hard to sift through it all and use it effectively, but machine learning has the potential to change that.

“One tool for working with potentially valuable truck loads [of data] is an artificial neural network — a software system that uses machine learning techniques to process tons of data and intelligently answer questions,” Ars Technica reported.

OmniEarth, a geographic analytics company, is using IBM Watson’s machine learning technology to give meaning to water data. The project has deployed this technology to help water managers understand whether ratepayers are using water efficiently. The technology analyzes “aerial and satellite photos to estimate the demand for water on a property-by-property level based on what the property contains,” Ars Technica reported.

Tom Ash of the Inland Empire Water Utilities Agency, a waste treatment agency and wholesale water distributor serving customers near Los Angeles, discussed the benefits of this approach.

“They can tell you how good, bad, or ugly customers’ water use is, how efficient it is, who’s wasting water, where that is,” he said, per the report. “And then [utilities] know who to target with what kind of program. If you don’t have any landscape then you target them for, jeez, maybe you’ve got leaks, maybe you’ve got high-flow plumbing devices that you need to retrofit — great, we’ve got a program. So it really helps make the efficiency level and outreach of your conservation programs much more effective.”

IBM’s Jerome Pesenti discussed the technological approach of this project.

“Our service itself has never really been trained for aerial imagery. It’s really about a general visual recognition service, but what they found is that it actually worked pretty well when they gave it enough training examples,” he said, per Ars Technica. “So then they took hundreds of thousands of images of the whole state of California, passed them through the classifier that they had trained, and they were able to identify all these features.”

OmniEarth Chief Strategy Officer Jonathan Fentzke weighed in as well.

“What we do is, for every parcel or region of interest, we calculate the square footage of tree and shrub and grass and pool and roof and irrigated and non-irrigated surface,” he said.

The engineering firm Black & Veatch cited the potential for data to improve water utility operations in its most recent report on the state of the industry.

There are “opportunities to deploy technology in the form of sensors and data analytics platforms to create value across the entire lifecycle of the utility system while improving overall business operations and safety of water utility providers,” the report said.

Analysts say water managers are missing out on opportunities for to use data effectively. Charles Fishman, author of The Big Thirst: The Secret Life and Turbulent Future of Water, laid out this argument with regard to federal water data in a New York Times editorial.

The federal water data system “was an embarrassment even two decades ago. The vast gaps — we start out missing 80 percent of the picture — mean that from one side of the continent to the other, we’re making decisions blindly,” he wrote.

 

For the latest news and conversations about AI in business, follow us on Twitter @Business_AI and join us on LinkedIn – AI Business Community

Source: http://bit.ly/29TgVDb

Feature image credit: Flickr

Leave a Reply

Translate »

We use cookies. By browsing our site you agree to our use of cookies.Accept