Author: Deepak Battula

Are Hybrid Body Types More Accident-Prone?

The Spark of Curiosity

I always wondered, are hybrid body cars more accident-prone than a traditional body style vehicle?

When I got to know about the crash of Rishab Pant, a famous Cricketer from India, I was devastated by the news. But when I saw the car he crashed in (a Mercedes-Benz AMG GLE Coupe), it planted a seed of curiosity in my mind. Are hybrid body cars more accident-prone than a regular body?

It's 4 AM on a Sunday, and I'm stuck with this question again on a sleepless night. I decided to find the answer myself. I've researched and checked with the data, but the Indian car market is dominated by a lot of crossover SUVs and hatchbacks. The USA has a more vibrant car culture and more available data, so I've chosen to use USA's data. I think this project would suit my passion for Cars, Engineering, and Data Science.

The Theory: My Take on 'Hybrid Bodies'

First, let's be clear. When I say "hybrid," I'm talking about the body style, not a plug-in hybrid that runs on fuel and electricity. A hybrid body type is a blend of two styles, like the SUV-Coupe, for example, the BMW X4.

Now, here's where my personal opinion comes in, and it's the whole reason I'm doing this project.

I've always held a strong opinion that these kinds of hybrid body types shouldn't exist. My theory is that when you create a hybrid of an SUV and a coupe, you get the worst of both worlds for safety. You're taking a vehicle with a high center of gravity (the SUV part) and marketing it for high-speed performance (the coupe part). This encourages people to drive faster, but the vehicle's physics can't fully support it like a true coupe can. This, I believe, makes the margin of error dangerously thin.

But that's just a theory. The goal of this project is to see if the data agrees with my gut feeling.

The Challenge: How to Measure "Accident-Prone" Fairly

So, how will I find out what cars crash more? It depends on so many factors, like the driver, speed, tire health, traffic, and many more.

A coupe or supercar will naturally have fewer total crashes than a common SUV or sedan. That's because coupes are less common on the roads, and people might not drive them as often. So, just counting crashes is the wrong approach.

What's the correct parameter? The key is to create a level playing field. To do this, I'll need to look at a combination of data:

The market share of each body type.
How many miles are driven per year, or Vehicle Miles Traveled (VMT).
The total number of crashes.
The number of fatal crashes.

By comparing the number of crashes to how much these cars are actually on the road (VMT), I can calculate a much fairer crash rate.

The Game Plan: My Tasks

Here's the checklist of what I plan to do:

Build a Tool: Eventually, I want a user to be able to input a car body type and get crash estimations and details.
Visualize the Data: Create charts and graphs to make the findings easy to understand.
Calculate Crash Rates: Compare body types against their crash rates.
Analyze Survival Rates: Look at the ratio of total crashes to fatal crashes for each category.
Answer the Big Question: Finally, determine if hybrid body types are statistically more accident-prone.

Keeping It Real: Project Limitations

I know this project can't be perfect. There are some things I just can't measure:

I can't consider if a driver was reckless, sleepy, or distracted.
A supercar is almost always driven more aggressively than other body types.
I can't account for vehicle health, like worn-out tires.
In collisions, sometimes innocent people are hurt, and it's hard to determine who was at fault from the data.
For now, I don't intend to add the time of the crash (day/night).
For simplicity, I'll be classifying Crossover SUVs under the general "SUV" category.

The Data Hunt: Where I'm Looking

I've found a few promising sources for this data. The big question is if it's legal to use for a personal project (I believe so, as it's public data, but I'll be careful). Here are the sources I'm considering:

National Highway Traffic Safety Administration (NHTSA)
Insurance Institute for Highway Safety (IIHS)
U.S. Department of Transportation - Federal Highway Administration (FHWA)
MarkLines Co., Ltd.
GoodCarBadCar.net (Automotive Sales Data Aggregator)
For future comparison: Ministry of Road Transport and Highways (MoRTH) & National Crime Records Bureau (NCRB) from India.

This is just the beginning. I plan to write a series of posts as I work through this project. Follow along to see what I find!