View Source: Brian Straight

Samsara data project finds that risky driving behaviors are more likely to occur early in a shift or near the end of a driving shift.

Commercial vehicle drivers are more likely to engage in unsafe driving behaviors at the beginning or end of their shifts. Meanwhile, U.S. drivers are speeding more frequently on the less-congested roadways during the COVID-19 pandemic.

These are some of the findings from a new data deep dive from Internet of Things (IoT) provider Samsara, analyzing data from its commercial customers.

According to Ali Akhtar, director of data science for the firm, reduced roadway congestion has led to a 20% increase in severe speeding among commercial drivers. In a blog post, Akhtar explained why Samsara took on this project.

“To help fleet managers better understand patterns in risky driving behavior, we wanted to know: how does driver behavior change over the course of a shift? When are drivers most likely to exhibit risky behaviors? And how can fleet managers use this data to coach drivers more effectively?” he wrote.

How Samsara analyzed the data 

Samsara sampled its datasets going back to Jan. 1. The data includes more than 2 million incidents of high-risk behavior obtained through g-force measurements and artificial intelligence data gathering.

The company joined trips data with its safety events database for four events: harsh acceleration, harsh braking, distracted driving, and tailgating. It also compared the overall data findings with subsets, including a statistically significant sample of individual drivers, and the relative occurrence of events against the top 40 U.S. cities by population. It then presented the results by breaking down the shift length into tenths and comparing the first and last tenth to the middle eight-tenths.

The data showed that unsafe driving behaviors such as harsh acceleration, harsh braking, distracted driving, and tailgating were 26% more likely to occur in the first tenth of a shift and 41% more likely to occur in the last tenth of the shift.

Samsara looked at shifts ranging from two to 12 hours in length.

“Despite the variation in shift length, our data shows a consistent trend across shifts,” Akhtar wrote. “When we looked at our sample of common unsafe driving behaviors, we found that they occur more frequently at the beginning and end of drivers’ shifts.”

Breaking down the data further, Samsara found that harsh acceleration was 77% more likely to occur in the final tenth of a shift while tailgating was just 16% more likely. Harsh braking was 54% more likely and distracted driving was 18% as likely to occur.

City driving may play a role in the trends, Akhtar wrote. For instance, distracted driving was 26% more likely to occur near the end of a shift that ended in a city versus 17% more likely when that shift didn’t end in a city.

Akhtar said it is impossible to know all the possible factors that contribute to the trends, but cities, last mile stops and driver fatigue likely play roles. He noted that while the Federal Motor Carrier Safety Administration has reported fatigue as a factor in 13% of large truck accidents, the length of the shift – when fatigue would seem mostly likely to become a factor – did not matter.

“While we didn’t control for driver fatigue in this analysis, we found that the trend persists in shorter shifts (two to four hours in length), when drivers are less likely to be fatigued,” Akhtar wrote.

Akhtar said that driver coaching can improve these behaviors.

“Our data shows that when drivers receive in-cab alerts for harsh braking and harsh acceleration, the frequency of those behaviors is reduced by up to 40%,” he wrote.

More on the results and methodology can be found here.