This morning when you filled the truck with fuel, something amazing happened behind the scenes. As you swiped your credit card in the reader, it triggered a series of events that determined:
- whether you had any credit left on the card, and how much
- if you were in a geographic location where the card is often used
- if you were buying something you’ve bought in the past
- if the purchase is typical for you
- if the zip code or PIN you entered matches the one assigned to your account.
A millisecond later, after all the inputs were analyzed, a datum center hundreds of miles away decided you were the card owner, and that you were eligible to charge the purchase. That everyday example of predictive analytics shows how more and more businesses are making better decisions by finding new value in the data they collect.
It’s Not Magic
Predictive analytics sounds complicated, and behind the scenes it is, but real world uses are easy to understand, and they make good business sense. And for construction companies, predictive analytics are offering the next step on the journey to incident-free jobsites.
Imagine if you knew that a particular task, done the way you always do it, under specific conditions, increases the chance of injury by 200 percent? Or, suppose a particular tool, used by carpenters when an alternative tool isn’t available, results in shoulder injuries three out of five times. Figuring out a way to reduce these injuries would be no small success when you’re talking about the wellbeing of your teams.
It Augments Human Thinking
Humans are very good at analyzing and drawing conclusions based on small collections of data. You can look at the prices and lengths of 2x4s from four different suppliers, and easily decide which is the best value for the application at hand. But, try doing that when you also have to consider the relative moisture content of 100 randomly sampled pieces taken from four different kilns where they completed the drying process on four different days, and traveled to the store on four different types of transportation. It’s doubtful you’ll get that analysis done without the help of some computing power.
So, predictive analytics work behind-the-scenes to give you a better chance of making the right decision because it uses a huge set of data. Not only does it help in making the right decision, but it also helps you to make it faster, and with greater precision. And, when it comes to safety––better, faster, and more precise wins every time.
Narrow, But Impressive Results
But, if predictive analytics work for safety, where is the proof? Researchers at Carnegie Mellon University studied four years of real workplace safety data and built predictive models that could predict the number of injuries a jobsite will have with 80 to 97 percent accuracy rates!
Another company, Deloitte, an audit and consulting firm, markets a safety analysis service called Smart. This product helps companies source, optimize, and analyze multiple factors that lead to workplace safety incidents. A mining company, Goldcorp. Inc., reported a 75% drop in injury frequency rates over a six-year period by using analytics services provided by Deloitte.
Besides collecting and analyzing data, companies are also turning to smart devices to predict potential safety issues. Caterpillar offers its own device, called a Cat Smartband, to alert wearers and managers when operators are lacking sleep, and are prone to fatigue. Some Cat customers are combining the wristbands with dash-mounted devices to detect severe braking, swerving, and other errant vehicle movement. Over time, the devices cause operators to become more aware of when they are becoming too inattentive to operate the equipment safely.
The Path to Getting Predictive About Safety
A prerequisite to harnessing predictive analytics for safety is to understand that analyzing historical accident reports only gives you what are called “lagging indicators.” Instead, you have to focus on the factors that cause incidents, and those types of data are not likely in your incident, or accident reports. Much of it is the kind that is only now becoming available through smart devices.
These types of data include:
- Vehicle and human telemetry
- Geospatial data
- Equipment telematics
But, there is also data hidden in your everyday business transactions that offer important insights. Information related to payroll and performance, social and demographic data, and training records are needed to make the safety picture even more accurate. Ultimately, you’ll be combining, collating, and sorting information from four spheres:
- Human Resources
Unless you have your own information technology talent on board, this is where you will no doubt need some outside help to really get rolling with predictive analytics for safety. But, once you get started, and you see the insights you get from your safety dashboard, you will see that you are taking accident prevention to the next level.
Besides seeing safety issues ahead of time, you will see where your preventive efforts are paying dividends. There will be less pain and suffering on your jobs, you will be more likely to avoid safety related fines, you will have a positive impact on your insurance premiums, your claims costs will be lower, human resources costs will be lower, and you will avoid litigation. And, you’ll do all of that before you stop to fill the truck in the morning.