This is a special research report on using big data for road safety. Jorge Gonzalez used Geotab telematics big data to analyze driving behaviour near accident black spots in Spain.
Hundreds of thousands of people across the world lose their lives in car accidents and road disasters every year. More than one thousand of those accidents happen on Spanish roads.1 In addition to the social impact, these accidents also have a huge economic impact. The estimated cost of yearly traffic accidents in Spain is 177 million euros and about 2 million lost working days.
The figure of 177 million doesn’t take into account insurance and compensation, medical costs and health resources, legal defenses, surveys, or decreased productivity in a company, among others. In 2015, there were 54,416 lost work days caused by traffic accidents — that’s 11% of work casualties in total.2,3
According to the World Health Organization (WHO), the economic cost of traffic accidents around the world is about US$ 518 billion.
It’s estimated that traffic accidents cost between 1% and 3% of the gross domestic product (GDP) of each country. The report titled Seguridad vial laboral: Una inversion rentable [Job road safety: A profitable investment] published by the European Transport Safety Council (ETSC) in collaboration with the Fundación MAPFRE, points out that road safety management provides an opportunity to reduce these costs in various ways and shows several practical examples.4
Among the major causes of traffic accidents listed by MAPFRE, the human element is the cause of 80% to 90% of these accidents.
The three major human failures, in order, are: distractions, speeding and alcohol. Failure to use a seat belt and manoeuvres such as harsh cornering, harsh acceleration, and hard braking are also high risk behaviours on the roads.5
Analyzing Driving behaviour in Accident Black Spots with Big Data
Research Goals: In Spain, “puntos negros” — or black spots — refer to sections of the road where there is a higher concentration of accidents. Black spots are those locations in a “road network in which during a calendar year three or more accidents were detected with victims with a maximum separation between each other of 100 m.”6
The main objective of this project was to design a big data framework suitable for the analysis of information collected by Geotab telematics devices installed in vehicles. The idea is to use this framework for future research on global safety. This study constitutes a first approach through a real use case analyzing the driving behaviour near accident black spots in Spain.
The analysis was made using Geotab telematics data in an anonymous and aggregated way. That data includes driving behaviour and GPS positions.
The Geotab GO7 device provides GPS and accelerometer data among other specifics. With this data, driving speed and any harsh driving events (accelerating, braking or cornering) can be identified and analyzed. Speeding, harsh driving and unbuckled drivers were considered high risk driving events.
Geotab devices generate a large amount of data, collecting over 2 billion data points per day, making the implementation of big data techniques a necessity for this project. The main technologies used were Google BigQuery and Google cloud Datalab.
The GPS data from Geotab was integrated with the latest black spot information provided by the Dirección General de Trafićo (DGT). Thanks to this integration, it was possible to search information about driving behaviour when the vehicles were in those black spots.
The black spots were grouped into two categories:
- Number of victims
- Predominant accident type
Once the information was properly structured, the amount of data and high risk driving events, as well as the type of high risk events for each category were analyzed. The results of this analysis have provided valued insight into the relationship between black spots and driving behaviour.
Results and Conclusion: Speed Causes Victims
There are two types of black spot categories, one depending on the number of victims in a year and the other depending on the predominant type of accident in that black spot. An analysis was carried out for each category.
Analysis 1. Comparison of Number of Victims with High Risk Driving Events
The table below shows the number of events/incidents with the number of GPS records near that black spot per category.
Table 1. Black spot incidents and GPS records by number of victims category
||Number of Victims in a Year
||Number of Incidents
||Number of GPS Records
||Number of GPS Records
||Less than 10
||Between 10 and 20
||More than 20
The resulting percentages did not follow any rule, so a comparison of the incident distribution per category was made.
Figure 1. Percent of high risk driving incidents by number of victims category
Speed is correlated with the gravity of the injuries related to an accident. When the number of accident victims is higher, there is approximately 5% more speeding compared with others incidents.
Further research is needed detailing why incidents of speeding increase in those black spots where there are more victims and what measures can be taken to reduce the number of victims.
Analysis 2. Comparison of Predominant Type of Accident with High Risk Driving Events
For this project, a black spot has a predominant type of accident if half or more of the accidents in that black spot are of that type. The following table compares the number of incidents with the number of GPS records.
Table 2. Black spot GPS records and incidents per accident type categories
||Number of Incidents
||Predominant Type of Accident
||out of road
Vehicle overturn stands out as the top type of accident at 42%. Collision has the majority of GPS records, thus the collision accident type will set the trend for incident distribution for each category.
Figure 2. Percent of high risk driving events for collision type
Figure 3. Percent of high risk driving for out of road accidents
Figure 4. Percent of high risk driving events for collision accidents
Looking at out of road accidents at black spots, the percentage of speeding increases conspicuously to 80.7%. The out of road category of accidents also has a high percentage of incidents — 20% (see table above).
Speeding or driving too fast for road conditions is the main cause of out of road accidents in this analysis.
Measures should be taken to decrease the number of speeding-related incidents. This can be done by investigating each out of road black spot and reviewing if the posted road speed is appropriate or if there is inadequate road signage indicating that it is a dangerous spot.
On black spots where run over is the predominant type of accident, the analysis shows there are fewer speeding incidents. The percentage of harsh cornering is double the average and the percentage of harsh braking/cornering increase from 10.3% to 15.3%. A sharp turn of the road with low visibility could be a possible cause of increase in harsh cornering and danger of run overs. Improving visibility or posting a warning might help increase the safety at this dangerous spot.
Figure 5. Percent of high risk driving events for other accidents
Figure 6. Percent of high risk driving events for overturn accidents
The distribution of the others type of accidents stayed similar to the averages.
For overturn accidents, the percentage of harsh cornering is much greater than the average for all the records: from 23% to 83.6%. Also, there is an increase in harsh braking/accelerating exceptions and a decrease of speeding from 60% to 0%. These results point out that the black spot is on a curve or a roundabout. Knowing that there is only one overturn black spot and that the percentage of incidents compared to GPS records is 42%, further analysis was required.
Figure 7. Map image of overturn black spot
As expected, the black spot is on a roundabout with a speed limit of 40 km/h. For this project, the speeding has been calculated detecting speeds higher than 90 km/h (maximum speed on Spanish roads for vans) because most of the vehicles analyzed were vans. The speeding calculation on this point was adjusted to 40 km/h, obtaining the following results:
Figure 8. Speeding, harsh cornering and harsh braking/accelerating with adjusted speed calculation
Changing directions inside the roundabout and getting out of it could be causes of the high percentage of harsh cornering. Adjusting the speed to get in or out of the roundabout could be an interpretation for the reason of the harsh braking and accelerating. Despite the new speeding rule, the speeding percentage is about four times smaller than the averages. One of the main causes of overturn accidents is a harsh cornering event. Adding a traffic signal or warning signage could help reduce those accidents.
For all the categories, the incidents that cause that type of accidents are the most frequent. This analysis proves that to improve road safety, the adjustment of signalization and the increase of awareness on black spots should be taken seriously. For example, if in the overturn black spot the incidents percentage is 42% and most of that incidents are harsh cornering, it is normal to have some overturn accidents.
Adding signalization or refurbishing the roads on the black spots according to its number of victims and type of predominant accident can prevent more accidents on those points.
Continuing with this project, future research should study each black spot and adjust the analysis to their characteristics.
The position of speed cameras or radars and the behaviour before and after them would be another interesting future line of work. It would be interesting to identify harsh braking events or speeding in those points in order to see if the cameras are needed there or on the contrary, if they are dangerous.
Another area of future research would be searching for potential black spots, analyzing areas of the roads in Spain where driving behaviour or incidents are worse than in the black spots by comparing the driving patterns.
This project shows the power of using big data and telematics to improve upon safety and the quality and quantity of data that is possible to process with the right tools.
Only the data of a few thousand vehicles have been analyzed in this project. This makes the possibility of redoing the analysis using the full dataset from the 750,000+ Geotab-connected vehicles, or simply by adding more information about black spots in other countries, even more intriguing and informative. If this information was not available, the driving patterns near the Spanish black spots could be compared with the driving in any point around the world. This project has been designed to allow for replication or expansion with changes.
Find more of our latest white papers, including a report on Smart Cities, on our Geotab Reads page.
- DGT, “Puntos Negros,” n.d. Retrieved from: http://www.dgt.es/es/el-trafico/puntos-negros/
- “Coste laboral anual de las bajas por accidente de tráfico es de 177 millones.” Expansión, Mar 8, 2016. Retrieved from: http://www.expansion.com/agencia/efe/2016/03/08/21736149.html
- “Las bajas laborales por accidente de tráfico ‘nos’ cuestan 175 millones.” El Mundo, Mar 8, 2016. Retrieved from: http://www.elmundo.es/motor/2016/03/08/56deb8e022601dcc698b45fa.html
- Fundación MAPFRE, “El coste económico de los accidentes laborales de tráfico,” Aug 24, 2015. Retrieved from: http://www.seguridadvialenlaempresa.com/seguridad-empresas/actualidad/noticias/coste-economico-accidentes-laborales-trafico.jsp
- Canales MAPFRE, “Las 5 causas principales en los accidentes de tráfico,” May 9, 2016. Retrieved from: https://www.motor.mapfre.es/consejos-practicos/seguridad-vial/6068/5-causas-accidentes-trafico
- Por Marcial. “Seguridad vial: Puntos negros.” Impenor, Jul 2, 2013. Retrieved from: http://www.inpenor.com/2013/07/02/seguridad-vial-puntos-negros/
Geotab Data Shows Impact of Hurricane Harvey
Predicting Traffic Congestion with Driving Behavior
How Fleet Safety Standards can Reduce Fatalities