I have a database about claims at auto insurance. I can do the math and get Frequency, Severity, Sinistrality. I need to know if cars with automatic transmission have a higher propensity to accidents than cars with manual transmission.
Some ideas I have:
- Comparing frecuencies among transmission and coverages (descriptive)
- An hypothesis testing in order to know if there's a difference between frequencies (as populations, but im not sure what kind of hypothesis will be appropriate)
- A logistic regression, where Y=1 if the car had a claim, 0 if not. Check if "transmission" has a parameter that differs from zero.
Any suggestion? am I right with my ideas?