We have a machine learning model that classifies a driver in different classes of risk based on their telematics data. I wonder what type of variable the risk is ?
As of now, if I have a car and have an insurance with a particular company - the premium is fixed for a month. So we know how much we need to pay per month, we can adjust and calculate the monthly expenses accordingly. In this case as explained - we won't have much idea how much I need to pay per month - rather than it will depend on where I am driving, is it peak hours or an empty road etc. What is the advantage in that case?
Good question. Perhaps you can have both options. Some utility companies are offering consistent payments throughout the year so you don't have to guess. That has its advantages and disadvantage is. Now the question of pricing and business models have to be taken into account also.
Payment can be fixed and adjusted periodically. Imagine having a balance with your insurer where, depending on your usage, you will have to pay an additional lump sum or get a refund for any difference between your estimated monthly premium vs. your real-time premium.
Thanks for watching a video created eight years back! That was my vision/idea back then, and we are slowly getting there. Industries and regulations continue to change as well....:-) Check out the more recent information on that topic from this year - 2024 www.insurancethoughtleadership.com/ai-machine-learning/how-ai-could-set-premiums-real-time www.simplesolve.com/blog/dynamic-pricing-in-insurance-using-ai And there's even an ML product on pricing for P&C insurance www.lumnion.com/insights/harnessing-machine-learning-for-next-gen-insurance-pricing-insights-by-lumnion
![[안방1열 풀캠4K] 베이비몬스터 'DRIP' (BABYMONSTER FullCam)│@SBS Inkigayo 241110](http://i.ytimg.com/vi/AQ4RIghEvUs/mqdefault.jpg)
This is a great video with a great explanation. Thanks for making this!
thank you
Highly educative videos! Hats-off to you Raj Sir
Thanks very much
great insight! Where can l find more about this subject?
informative and valuable insight! Thanks.
We have a machine learning model that classifies a driver in different classes of risk based on their telematics data. I wonder what type of variable the risk is ?
risk could be the output variable. Then you map risk to premiums
Very well explained sir!
Great video, quite easy to follow.
As of now, if I have a car and have an insurance with a particular company - the premium is fixed for a month. So we know how much we need to pay per month, we can adjust and calculate the monthly expenses accordingly. In this case as explained - we won't have much idea how much I need to pay per month - rather than it will depend on where I am driving, is it peak hours or an empty road etc. What is the advantage in that case?
Good question. Perhaps you can have both options. Some utility companies are offering consistent payments throughout the year so you don't have to guess. That has its advantages and disadvantage is. Now the question of pricing and business models have to be taken into account also.
Payment can be fixed and adjusted periodically. Imagine having a balance with your insurer where, depending on your usage, you will have to pay an additional lump sum or get a refund for any difference between your estimated monthly premium vs. your real-time premium.
This will never work in practice and you obviously don't have any experience with actuarial topics or even regulation of the industry.
Thanks for watching a video created eight years back! That was my vision/idea back then, and we are slowly getting there. Industries and regulations continue to change as well....:-)
Check out the more recent information on that topic from this year - 2024
www.insurancethoughtleadership.com/ai-machine-learning/how-ai-could-set-premiums-real-time
www.simplesolve.com/blog/dynamic-pricing-in-insurance-using-ai
And there's even an ML product on pricing for P&C insurance
www.lumnion.com/insights/harnessing-machine-learning-for-next-gen-insurance-pricing-insights-by-lumnion
Very vague video
Yeahh stilll more to add on. Real life implementation is gonna be much more hectic thn this. wht says?