2

Hi i am actually not an actuary but I'm currently working on an ML project on insurance pricing. The pricing is calculated by multiplying Claim Frequency and Severity, correct?

I'm currently using Poisson GLM to predict the Claim Frequency, using Exposure as the offset, the data I have in hand is historical claim data of policyholders from 2019 until now.

My question is, can I use the multiple calendar years as the exposure for the Poisson glm, or do I need to train the data on a yearly basis? Since all the research papers I read on this use only 1 year of data for the models.

(Sorry for my bad english and the noob question lol)

kjetil b halvorsen
  • 77,844
Fachri Ardiansyah
  • 21
  • 1
    Did the papers say why they analyze only one year of data at a time? Maybe because the large sample size permits it ... anyhow, there is no mathematical or statistical reason not to use yearly exposure as an offset, also see https://stats.stackexchange.com/questions/373890/separate-models-vs-flags-in-the-same-model/373909#373909 – kjetil b halvorsen Dec 21 '23 at 16:28

0 Answers0