July 19, 2024

[ad_1]

Vehicle telematics data can reveal a lot about your driving skills

Photo of myself driving to collect vehicle telematic data (image by author)

Are you a good driver? The answer to such a question has not been very objective. One way to analyze this is to take the opinion of the passengers traveling with you or simply count the speeding tickets you paid! However, these are all very primitive ways to judge driving behavior. In this blog, I will show on how I have used data to analyze my driving behavior objectively.

Driving behavior analysis is also used by enterprises for use cases such as designing a data-driven insurance policy or vehicle fleet management.

In this blog, I will go through data science techniques that can help measure driving behavior such as

  • Over speeding
  • Hard Acceleration
  • Anticipation
  • Machine learning to see if bad driving impacts vehicle conditions

The first thing to analyze how one drives is to collect data. Most vehicles have sensors that measure various things such as speed, temperature, acceleration, and many more. This is done using a Vehicle telematic device. There are various vendors which provide such devices.

Vehicle telematics device (image by the author of device installed in the car)

This device takes data recorded by sensors and then transmits it to the vendor database. The data can then be used to understand driving behavior. You can also ask your vendor to provide access to the data so that you can analyze it in more detail.

Vehicle telematics data collection (image by author)

In this blog, I will use an example of data that I collected during my trip to India in the state of Karnataka. The data collected is for the 21st of August 2022. We will attempt to find out if my driving is good or not. The data which is collected has information as shown below.

Sample data collected from vehicle telematic device (image by author)

The data has a device ID that identifies the telematic device. It has the timestamp of the data record, as well as various things, that get measured, for example, the position of the vehicle which is measured as the latitude, longitude, and elevation. The speed of the vehicle is measured in KMPH or MPH

Let us now analyze driving behavior.

Over-speeding is one of the first things which can be measured in order to understand driving behavior. Shown here is the route that I took on the 21st of August 2022. The data used to make this analysis is based on the telematic data on time-stamp, latitude, and longitude.

The trip is from Udipi to Holekattu. The route taken is highway number 66 which runs along the western coast of India.

Vehicle Trip visualization (image by author produced using Google Maps and Javascript)

You can also observe a marker, which is a location where the maximum speed of 92 KMPH was recorded. The speed limit for a car on national highway 66 is 100 KMPH. So the vehicle has been within speed limits, and we can give a green tick for speeding behavior.

Driving behavior for speeding is ok (image by author)

Hard acceleration is an event where more force than normal is applied to the vehicle’s accelerator or brake system. Some people may refer to this as ‘lead foot’ syndrome, and it can be an indicator of aggressive or unsafe driving behavior.

Let us now measure hard acceleration during my trip. Shown here are some of the other markers before the maximum speed of 92 was achieved, which shows the speed of 73, and then to 85, and then achieving a maximum speed of 92.

Markers before maximum speed (image by author produced using Google Maps and Javascript)

We can put these vehicle speeds in a time perspective with the line plot as shown below. You have the time on the X-axis ad vehicle speed on the Y-axis. This curve corresponds to acceleration. I started to accelerate at 14:43:21 when the speed was 71 and then reached the max speed of 92 at 14:43:49. So I increased my speed by 21 km/hr in 28 seconds.

Time vs Velocity (image by author)

In order to see if this acceleration is harsh acceleration or not, we need to convert it into gravitational force, also called g-force, which gets applied to the car due to acceleration. A speed increase of 21 km/hr in 28 seconds corresponds to a g force of 0.208 m/s2 acceleration. Shown below is a mapping between g-force to levels of acceleration.

Acceleration or braking mapped to gravitational force (g-force) (image by author)

A g-force from 0.28 is considered safe and is not hard acceleration. So, we can give a green tick for hard acceleration.

Driving behavior for hard acceleration is ok (image by author)

Anticipation in driving means reading your surroundings and remaining aware by keeping your eyes and ears open. It means planning well ahead and being prepared to take necessary action. To anticipate and plan for others’ actions you should be constantly checking what is happening all around you.

Let us now check on my anticipation skills. In order to see my anticipation skills, we can analyze what I did after acceleration. Looking at the curve shown below, we can observe that there is a sudden decrease in the speed.

Visualizing de-acceleration (image by author)

From a speed of 92KMPH, I de-accelerated to 1 KMPH in 24 seconds. This is equivalent to a g-force of — 0.3 and corresponds to hard braking. What could be the reason?

Let me reveal the secret, as I know exactly what happened as I was driving the car. If we observe the route after the maximum speed point, we see a river bridge called the Heroor bridge. This bridge had reduced speed limits, which was the reason I had to de-accelerate.

Visualizing de-acceleration (image by author produced using Google Maps and Javascript)

Reducing from a high speed of 92 KMPH to a very low speed of 1 KMPH is a clear indication that I did not anticipate. So let us give a red for anticipation!

Driving behavior for anticipation is not ok (image by author)

Let us now see if driving behavior impacts the vehicle or not?. The telematics device collects data related to any alarm raised by the vehicle. A zero indicates no issues, while a 1 indicates a problem with the vehicle.

There are also more than 50 sensor values, such as vehicle speed, acceleration, oxygen, throttle, air temperature, and many more.

We can use a machine-learning decision tree to find any relation between sensor values and alarms. This will help us know which of the factors impact vehicle health.

Using a decision tree to find the relation between sensor values and alarm (image by author)

Shown below is a decision tree that has different sensors as decision nodes and alarms as an output node. You can see that the top factors which lead to vehicle alarms are BATTERY, ACCELERATION, and SPEED.

So bad driving behavior not only impacts driver safety but also impacts the health of the vehicle.

Top factors impacting vehicle health (image by author)

So here are some interesting conclusions

  • Data collection using telematic devices is key to data-driven driving behavior analysis
  • Analyzing speeding requires you to integrate vehicle speed data with speed limit data
  • Hard acceleration and anticipation can be calculated using time series functions. However, they need to be put in perspective with route analysis
  • Bad Driving behavior is not safe for the driver as well as the vehicle

If you like my mini-project of analyzing my own driving behavior through data and using data science techniques, please join Medium with my referral link.

Please subscribe to stay informed whenever I release a new story.

You can visit my website to make analytics with zero coding. https://experiencedatascience.com

On the website, you can also participate in upcoming virtual workshops for an interesting and innovative data science and AI experience.

Here is a link to my YouTube channel
https://www.youtube.com/c/DataScienceDemonstrated

How I analyzed my own Driving Behavior using data science Republished from Source https://towardsdatascience.com/how-i-analyzed-my-own-driving-behavior-using-data-science-d3a33efae3ec?source=rss—-7f60cf5620c9—4 via https://towardsdatascience.com/feed

<!–

–>

[ad_2]

Source link