The Next Game Changer: Predictive Analytics

Transcription

1 Session No. 596 The Next Game Changer: Predictive Analytics Del Lisk, CTP Vice President, Safety Services Lytx. Inc. Introduction Traffic accidents and the tragic consequences impact everyone in the United States. Most people have had someone close to them killed or injured due to traffic a traffic accident at some point in their lifetime. As reported by the National Highway Traffic Safety Administration (NHTSA) in 2013 alone 32, 719 people perished in motor vehicle incidents in the U.S. Another 2.3 million people were injured. Total reported crashes were more than 5.6 million. These statistics shockingly demonstrate how dangerous driving can be. On a brighter note, both fatalities and injuries have dropped significantly over the years when normalized to miles driven. However, this positive trend has stalled in the past few years as results have flat lined as the following chart from NHTSA demonstrates.

2 Business Fleets Business fleets have a strong motivation to reduce traffic accidents. By reducing traffic accidents they are improving the safety and well-being of their employees as well as the motoring public in general. But reducing traffic collisions can also lower overall costs to operate the fleet. In fact, avoiding just one serious crash could save an organization millions in potential litigation costs. As reported by the National Safety Council in Estimating the Costs of Unintentional Injuries in 2012, the average cost of a death from a motor vehicle crash is more than $1.4 million dollars. Fleet managers undertake several efforts in an effort to improve employee driving behavior and lower crash frequency. Common efforts include: Pre-employment screening efforts such as motor vehicle records and other background checks as well as personality tests. Defensive driver training courses. Monitoring programs such as 800 How s My Driving? Periodic check rides or covert driving observations. More and more fleets have begun to adopt technologies to further their driver safety efforts. Technologies such as telematics, forward collision warning systems and lane departure warning systems are examples of this. Increasingly, fleets are also using video based feedback systems to enhance driver improvement and coaching efforts. The Federal Government The federal government is playing a more active role in the safety of fleets. The Federal Motor Carriers Safety Administration (FMCSA) developed the Compliance, Safety and Accountability (CSA) Program to improve large truck and bus safety and ultimately reduce crashes, injuries and fatalities that are related to commercial motor vehicles. Its goal is to not only provide a scorecard as to where each fleet ranks on safety but also to include a predictive element so fleets know where they stand as it relates to the probability of a collision. The program relies heavily on data from roadside inspections as well as DOT recordable incidents to populate its database. One of the key elements holding fleets and drivers accountable to the goals of CSA is law enforcement. But, with more than two million trucks and only 15, 000 inspectors on the road, it s impossible for this small army of inspectors to find everyone who is not meeting standards. According to the Centers for Disease Control, fatal vehicle collisions are one of the leading causes of death in the United States today with one person dying in a collision every 15 minutes. Over 90 percent of these collisions involve human error as determined in the Indiana Tri-Level Study. This is exactly why it s important that fleets leverage technologies that go beyond the law s reach. These technologies not only help to isolate drivers with dangerous habits so they can be coached for improvement but they also collect volumes of data that can be used to help predict who and what are the biggest risks on the road. This is information that fleet managers can use to intercede before dangerous patterns lead to tragedies.

3 Making Data Sense Of course, technology alone won t make our roads safer. It s the analysis of the data that the technology provides that will help fleets understand what s going on behind the wheel. Predictive analytics helps make sense of the volumes of data that fleet managers are receiving. Predictive analytics encompass a variety of techniques: statistics, modeling, machine learning and data mining. By analyzing current and historical facts, one can make predictions about future probable events. In business, predictive models identify patterns found in historical, transactional and predictive data to identify risks and opportunities. Predictive models seek to explain the relationships between a critical variable and a group of factors that help predict its outcome. There are certain situations where predictive models can be especially beneficial in delivering valuable insights: Processes that require a large number of similar decisions Situations with high consequences (e.g., lives, injuries or high dollars at stake) Enormous amounts of data that must be analyzed Areas where it s possible to insert a model calculation into the actual process, either to automate decisions or to support human decision-makers The growth of in-cab and mobile technologies means more objective data and an improved ability to develop predictive models to assess driving risk. In cab video with millions of behavioral observations meets all of the conditions above, and makes it an ideal platform to develop predictive models. In conjunction with video and observational analysis, it s possible to deliver very accurate predictive models to determine how drivers drive and the risk that is associated with their driving patterns. The Power of Video Using predictive analytics, it s now possible to predict who is more likely to be involved in a future collision. Based on our analysis, when looking at the same drivers and time periods, telematics-only solutions are somewhat limited as contextual data is lacking. Traditional telematics can isolate behavior such as hard braking, sudden swerves or aggressive driving to then provide a predictive model. But it has limitations as contextual information such as driver attentiveness, traffic density and weather are not available. The inclusion of in-cab video and subsequent analysis provides a much greater insight and thus a more accurate predictive model. Imagine two drivers, each with four hard braking events. Looking at the telematics only data these two drivers would appear to have equal risk if their drive time or miles driven was similar. Now imagine if we also had in-cab video associated with each hard braking event. In Driver #1 we see the following: All four events were in very light traffic Two events were rough stops due to being caught off guard by a changing traffic light

4 One was due to a hard brake to avoid a rabbit running across the road One event was a harsh brake when a merging vehicle unexpectedly cut across multiple lanes. Here s what we see when looking at the video associated with Driver #2: The events were all in moderate to heavy traffic Two were due to hard braking when Driver #2 was distracted by a cell phone and had to brake hard for slowing traffic ahead The other two hard braking events were triggered when Driver #2 was contesting another vehicle for lane position during a merge. Adding the context provided by the video analysis gives a much more accurate set of data to determine which driver has a greater probability of a traffic collision. Expanding the data by combining video with other data can significantly increase the predictive power of a black box telematics solution. In-cab video enables more accurate predictive analytics by providing the context to events. The Value of Predictive Analytics Most fleet operators are drowning in a sea of data. Information that may be available includes RPM data, over speed, excessive idling, hard braking events, headway warning, lane departures, hours of service and much more. With all this data it can be difficult to see the forest through the trees. That s the role of predictive analytics as it relates to managing fleet safety. The relevant data is collected and analyzed and presented in an easily understood form so managers can easily digest the data and see where best to put their focus.

5 Predictive analytics helps managers to identify where to focus driver safety efforts. Summary Fleet managers have been dependent upon lagging indicators for far too long in their efforts to improve driver safety. The advancements of predictive analytics through the addition of the contextual data captured through in-cab video are the next leg up in enhancing fleet safety. Safety professionals can have at their fingertips valuable, accurate information that will guide them to making more informed decisions about who requires training or other interventions and when it needs to occur. The result will be more effective and efficient safety programs that will enhance the safety of their employees.