A
AJL
Member
Hi
The EVT core reading seems to suggest that the block maxima approach involves splitting up data (e.g. claim values) into blocks of a fixed size (fixed number of claims in each block) and considering the maximum value in each block. However the R example in the core reading and PBOR materials are framed differently in considering blocks of a time period (e.g. blocks of 6 months) where the number of claims involved is then not fixed. Is this just to demonstrate that you can define blocks in different ways, or should blocks of claims generally be considered in time periods? I'm guessing either way they tend towards the GEV for a large number of blocks.
The EVT core reading seems to suggest that the block maxima approach involves splitting up data (e.g. claim values) into blocks of a fixed size (fixed number of claims in each block) and considering the maximum value in each block. However the R example in the core reading and PBOR materials are framed differently in considering blocks of a time period (e.g. blocks of 6 months) where the number of claims involved is then not fixed. Is this just to demonstrate that you can define blocks in different ways, or should blocks of claims generally be considered in time periods? I'm guessing either way they tend towards the GEV for a large number of blocks.