@@mdevans43 makes sense. Im writing a master thesis of XGboost for pure premium housing insurances in denmark. I have a problem with modelling the frequency, as a lot of the frequencies will be 0, due to no claims during the year. What would your approach be to model frequency? My thought is to model the frequency first, then the severity. And in the end multiply those for the pure premium.
Thanks for the video. One point should be mentioned: xgboost isn't using gradient descent internally so what you are saying about eta from 10:18 is not accurate.
nvm got it it has to do with the many trees model framework now i have to go read about the differences between an objective function in a model like this and a more simple ML model sigh thanks for the rabbit hole favi
Very informative and interesting.
Thank you so much both! Very informative 🙂
What are you modelling? What is the target variable? I did not catch it
The data is generated as discussed in the opening sections. The intention is that the target variable resembles losses from motor insurance.
@@mdevans43 makes sense. Im writing a master thesis of XGboost for pure premium housing insurances in denmark.
I have a problem with modelling the frequency, as a lot of the frequencies will be 0, due to no claims during the year.
What would your approach be to model frequency?
My thought is to model the frequency first, then the severity. And in the end multiply those for the pure premium.
Thanks for the video. One point should be mentioned: xgboost isn't using gradient descent internally so what you are saying about eta from 10:18 is not accurate.
any clue what the eta would be for then?
Just read eta is the learning rate applied after each boosting step. what's the difference between a boosting step and a gradient step
nvm got it it has to do with the many trees model framework now i have to go read about the differences between an objective function in a model like this and a more simple ML model sigh thanks for the rabbit hole favi