Data Fusion for Electric Utilities

     

Data Fusion for Electric Utilities.png

This will be the first in an occasional series about linking consumer databases to other, non-traditional data sources for deeper and innovative insights.  Today we pick on electric utilities.

Electric utilities are a rich source of raw resources for your data mining pursuits.  They offer many opportunities to link and consolidate disparate sources of data, creating a fertile sandbox in which the data scientists and analysts can play.  Smart meter data, of course, provides minute-by-minute reliability performance, and utilities use this information to manage their critical systems and business metrics.

But there are other data sources as well.  The key is to link individual consumer data to utility data in a logical, effective manner.  The very nature of electric service territories—as defined by the ways the various utilities carve up their states—enables this linkage quite nicely.

Few Americans have real choices when it comes to their electric providers, at least in terms of where their households’ electrons originate.  In some states, notably Texas and New York, consumers can choose from a number of providers for billing purposes, but generation and transmission come from the same places, like NRG or ConEd.  By and large, consumers’ electricity provider depends on where they live.

Generalizing about consumers’ lifestyles based on their electric providers is a risky business; not because the insights are not useful, but because comparisons between the customers of Provider A and those of Provider B are no more meaningful that basing those comparisons on some pre-defined geographic definitions.

Still, for purposes of drawing comparisons between customers of different electric utilities, using county by county definitions is a good proxy for ascribing consumers an electric utility, especially when a couple of utilities provide service to the lion’s share of a given state’s residents.  Two utilities—Consumers Energy and DTE—provide electric service to 92% of Michigan’s 3.8 million households.  Linking data about Michiganders’ electric provider based on their county, along with DDG’s DecisionPoints consumer database (which is at an individual adult level and includes county of residence), and other data sources offers all sorts of insights.  Some are more useful than others.

5 Slightly Useful and Potentially Entertaining Facts about Michiganders and Their Electric Companies

1 - DTE customers are happier than are Consumers Energy customers

In 2016, DTE was second only to MidAmerican Energy in JD Power satisfaction ratings for large utilities in the Midwest US.  Consumers Energy was in the middle of the pack by this measure.

2 - More DTE customers were born in January, and more Consumers Energy customers were born in March and September.

On average, 8.3% of people should be born in each of the 12 months of the year.  But oddly, some consumer data says otherwise.  Here is a histogram of months of birth for adults in Michigan generally.Chart 1.pngThere’s a spike in January and another in June.  These two anomalies are almost certainly due to some fancy imputation algorithm that assigns people with unknown birthdays to the first day of the month—assuring that in some cases, a nearly a year is deducted from one’s actual age.  A different algorithm assigns unknown birth months to the middle of the month, causing a 6-month “wrong month” exposure to some folks.  How do I know this to be the case?  I don’t.  But I know how propeller-heads think.  And I am a propeller-head.  QED.  You may ask, "of all of the 800+ variables available to you, why on earth did you choose month of birth?"  Because I can.

3 - DTE customers are more than twice as likely to live in a condo.

And of course they’re infinitely  more likely to live in Detroit, since Consumers Energy doesn’t service much of that area.  There are proportionally more condos in Detroit than there are, say, Traverse City.  Nearly 7% of DTE customers live in a condo, while 3% of Consumers Energy customers do.  Which leads directly to the fact that…

4 - Consumers Energy customers have smaller living spaces to cool.

In Michigan, cooling is rarely one’s major concern; heating, on the other hand, plays a more prominent role and is generally in the purview of the gas company, and not the electric company.  Despite the preponderance of condos in DTE’s service footprint, Consumers Energy customers’ homes average about 1,800 square feet, versus 2,000 for their DTE counterparts.

5 - Lifestyle and social media behaviors are different for the two groups of utility customers.

The DecisionPoints consumer database contains a couple hundred propensity models, assigning a likelihood for the relevant variables for each adult in the US.  No electric utility customer is more likely than any other to buy “green” products, although DTE customers are more likely to donate to environmental causes and to live in an environmentally-focused household. Consumers Energy customers are more active on Facebook and Pinterest than DTE customers, while DTE customers tend to gravitate toward Twitter.

Watch this space for more fun with data fusion applications.  The next article will look at data fusion at a level far more discrete than counties...a single address.


Learn More

Want to learn more? Want to talk to Dino about your needs and how research, data and analtyics can help. Request a call with Dino today.

Request a Call with Dino Fire

About The Author

Dino Fire

Dino Fire, Data Decision Group's Chief Data Science Officer, joined Data Decisions Group in 1997 after nearly ten years as a senior manager with The Arbitron Company. Dino manages a team of research analysts, and brings over 25 years’ experience in quantitative research methodology, advanced research analysis, predictive modeling analytics, and data science.

His specialization is the integration of survey data with transactional, customer-based, or third-party data to provide holistic views of consumer attitudes and behaviors. This practice helps our clients develop optimal products and marketing content, and provides insights that might otherwise go undiscovered.

Dino’s undergraduate degree is in political science with a concentration in statistics from Kent State University (1983). He earned a Master of Science in Predictive Analytics from Northwestern University in 2013.