Using Predictive Analytics to Improve Fleet Decisions
Photo courtesy of iStockPhoto.com.
Predictive analytics can be a powerful decision-making tool within various parts of fleet operations. It’s used in many major industries such as retail, finance and insurance to forecast what is most likely going to happen in the future — it’s a popular risk assessment tool. To illustrate the potential uses in the world of fleet, safety and safety-related decisions provide a great example of how predictive analytics can impact fleet operations.
Fatal vehicle collisions are one of the leading causes of death in the U.S., according to the Centers for Disease Control. While this may not come as much of a surprise considering these types of fatalities frequently make headlines, what should be more alarming is the fact that many of these collisions are preventable.
With the nature of fleet business often revolving around its employees being on the road, fleets are significantly impacted by these statistics. Automotive Fleet reports most company drivers average 20,000 miles per year, with more fleets experiencing an increase in preventable accidents. The primary cause of this uptick in preventable fleet accidents points to driver distraction that now contributes to the 25-30% of all fleet-related accidents, reports Automotive Fleet.
By the Federal Motor Carrier Safety Administration’s definition, a preventable accident is “one which occurs because the driver fails to act in a reasonably expected manner to prevent it.” The National Safety Council’s Safe Driver Award Program adds to this definition for fleets by classifying it as “any accident involving an organizational vehicle which results in property damage and/or personal injury, regardless of who was injured, what property was damaged, to what extent or where it occurred, in which the driver in question failed to exercise every reasonable precaution to prevent the accident.”
Unsafe driving practices such as talking or text-messaging on a cell phone, driving while drowsy or intoxicated, and speeding not only fall into this classification and increase the likelihood of a vehicle crash, but are all conscious efforts within the driver’s control.
This puts pressure — often from company HR, legal and risk management departments — on fleet managers and senior managers to control these accidents due to liability exposure. Some companies have even resorted to charging deductibles for these “at-fault” accidents in an attempt to negate the reckless driving behavior that leads to these budget-busting mishaps.
Overall, the total accident rate for commercial fleets averages 20%, sometimes even higher in industries such as pharmaceutical, according to Automotive Fleet surveys. Industry studies show that accidents typically represent 14% of a fleet’s total expenses, not including soft costs such as downtime and lost employee productivity, to name a few. Advances in new vehicle safety technology over the past decade have also resulted in hiked up costs in the range of several thousands of dollars to repair damage and/or replace parts. This represents a huge opportunity to cut fleet costs and, more importantly, avoid tragedies.
While it may not be possible to reduce all preventable accidents, reducing the number by half would yield substantial cost savings as well as fewer injuries and fatalities on the road. Modifying driver behavior is one solution to not only cut down on preventable accidents, but also the associated costs.
Since fleet managers have the responsibility of ensuring driver safety, to have the capability to not only monitor risky behaviors and improve them over time but to be able to predict a potential accident before it occurs and prevent it from happening would be paramount in making a significant reduction to this risk. This is when predictive analytics comes into play.
Through telematics and other data, predictive models make it possible for fleets to make more educated decisions with less expense — it provides support for decisions, making them more efficient and effective, or in some cases, can be used to automate an entire decision-making process. Considered the most revolutionary technological step of the “Big Data” era, predictive analytics has quickly become one of the most advanced forms of customized risk management.
Predictive analytics is ideal for risky driving analysis, for example. The challenge with identifying high-risk drivers is the most transparent records such as driving records, traffic violations, and accident reports, and may not always indicate a driver with the highest potential risk. Armed with information on driving behaviors through telematics data, companies can put drivers in safety training programs tailored to their risky driving behaviors before a collision occurs. Information about a driver’s behavior can also be used to determine how likely an individual is to be involved in an accident as well as the costs associated with that risk.
Outside of safety, by leveraging and analyzing data to provide insights into vehicle and equipment usage, driver behavior, and fleet productivity schedules, fleet managers can discover various areas where cost efficiency can be applied. This includes preventive maintenance, which can greatly contribute to cost cutting and increased productivity. Trending out maintenance data creates the opportunity to predict component failure, and provide real-time transactional repair costs, helping fleets to shape best-in-class maintenance policies and procedures.
Looking at information such as past usage, maintenance and overall total cost of ownership, fleets can also use predictive analytics for vehicle procurement strategies.
Predictive analytics can help in other ancillary ways as well. One fleet that implemented the driver safety score model in its operations noted an increase in personal communication, with drivers sharing information and essentially comparing notes and scores. After using the program for several months, the client reported drivers calling in to discuss safety proactively and in a positive way, whereas in the past, safety had only been a source of punishment for a driver when they made a mistake, so it was often a topic that drivers avoided. This truly changed how safety was viewed by the driver population and the organization as a whole.
Predictive models come in various forms, depending on the behavior or event they are predicting. Most generate a score, similar to a credit score, with a higher score indicating a higher likelihood of the given behavior or event occurring.
Predictive analytics encompasses a variety of techniques from statistics, modeling, machine learning and data mining that analyze current and historical facts to make predictions about future or otherwise unknown events.
Predictive analytics tells the customers, with as much accuracy as possible, what will happen in a few weeks or other chosen time frame — not what has happened in the past, but what the most likely outcome for a specific driver or vehicle is in the future and why that’s the case. Predictive analytics gives fleets an opportunity to think proactively rather than reactively.
Using safety as an example, by analyzing a driver’s particular driving style via telematics data, a safety report can be compiled highlighting the risks inherent to the driver and help coach safer driving habits. Drivers are classified into categories of risk based off the probability that a driver will be in a collision. By making these probabilities easy to assess, a safety score can help fleets make better decisions on how to prioritize risk within their operations. The primary objectives of a safety score would be to identify risky drivers prior to a collision to provide the driver an opportunity to modify those behaviors and prevent the collision from occurring.
Like other types of risk assessment programs, predictive analytics is not intended to be a magic bullet. It is only the beginning and requires a total commitment from the organization, fleet personnel, and driver to turn the model’s efficacy into reality. It requires strong management commitment in order to succeed. The most successful users are organizations that involve all departments in the process at an early stage including operations, safety, HR and IT. One reason previous predictive analytics projects have failed is due to the failure to achieve alignment and full enterprise adoption.
The main component in leveraging fleet analytics is assembling and analyzing actual fleet data. In order to develop accurate and valuable predictive models, it’s important to understand the challenges you want to address first to ensure that the models are solving a real-world problem. Real data and real problems are the key.
Start by choosing the area of greatest need (e.g., safety, maintenance, or workers compensation). Large volumes of data can result in hours of analysis with no real return in the way of fleet operational and cost optimization. One step to curb data overload is to simply ensure the data accumulated is the data actually needed. Data and analytical models should align with overall fleet goals and provide measurable and meaningful results. One good place to start leveraging analytical data is to understand your organization’s goals beyond fleet as well. Being able to meet these goals through the use of measurable analytics is the objective.
Factors that drive a company’s ability to derive insights into risk include:
Next, identify the key elements and the type of data that will make up your fleet’s predictive model. The necessary data is then extracted, which should be validated by the fleet. Analytics involves more than just having the data — fleets must ensure they understand what the data is telling them and why it will help them predict a future event.
The best way to balance the accuracy and timeliness of the predictive model, such as the driver safety score described earlier, is to run the models every few weeks or the time frame that makes the most sense for your operations. Giving it at least a few weeks provides safety and operational groups enough time to take action, while still accurately reflecting the changes that indicate risk. This is balanced with trying to minimize operational burden since most have limited time, whether it’s coaching drivers or needing to get a vehicle in for repair before a major breakdown.
Photo courtesy of iStockPhoto.com.
To further the safety example, the program’s success could be measured by the percentage of drivers involved in collisions that were predicted accurately in the 90-day period prior to the collision. More specifically, it would examine the data of the risky driver to determine if that individual was classified at the highest risk level at least one-third of the time in the 90 days prior. This example uses an odds ratio, which is the measure of association between an exposure and an outcome. To explain further, it looks at the odds a driver will get into a collision in the next 90 days given a particular score.
Although predictive analytics is most commonly used with fleets who were early adopters of telematics, it is something that more and more fleets are seeing the value in. With predictive analytics, fleets can proactively take action and make educated decisions that positively impact their organizations by reducing risk, whether tied to improving safety and costs or decreasing the work behind the decision-making process.