0
$\begingroup$

I have two classification models: The first model classifies driving behaviors to normal and aggressive; the output looks like this [0.91,0.09]. the second model classifies driving patterns to (speed-up, slow-down, Turn, steady), the output looks like this [0.42,0.38, 0.15, 0.05]. I want to use these outputs to give the driver a score (0 to 10) over a period of time (for example 20 minutes), knowing that the classification is done every 60 second (each 60 seconds have a pair of classes, ex (Aggressive, Turn)). is there any algorithm to do this?

I tried to use fuzzy logic expert system, but the output can't be fuzzified. thanks for your help.

$\endgroup$

2 Answers 2

0
$\begingroup$

You will need to create a formula to take your inputs that calculates a driver score. That formula will look like whatever you want it to because you are the one defining what makes a high and low driver score.

Start by thinking what makes a good driving score and what makes a bad driving score. Which variables correspond to that and how do the other variables change those. There is no standard formula. You are the one defining what it means.

$\endgroup$
1
  • $\begingroup$ Thanks very much for your help $\endgroup$
    – karam dar
    Commented Sep 7, 2023 at 13:09
0
$\begingroup$

Think simple, do you have a dataset with driving scores? If you do you can use supervised learning algorithms. If you don't, and there is no good way to create a dataset for such a large input-output space, maybe think out of the box.

You can create a simulation environment and go with reinforcement learning algorithms, then you can train both models with the reward functions from the environment.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.