I use software in my work called Emblem to perform GLIMS. When fitting the model the software uses the iteratively reweighted least squares algorithm. I am interested in understanding how this works in detail and I would like to do a few iterations using a very basic data set. I know the formula is available online but it would be really helpful if there was an example of this somewhere i.e. where the first couple of iterations are illustrated using actualy numbers.
Below is the dataset I am using. It contains only two rating factors and I would like to perform the algorithm on this dataset. It's a claim frequency model so the distribution is assumed to be Poisson.
Policy years Claims Vehicle Size Vehicle Type
500 42 small 1
1200 37 medium 1
100 1 large 1
400 101 small 2
500 73 medium 2
300 14 large 2
Does anyone know of a resource that might illustrate the approach to the algorithm using a worked example and a couple of iterations to illustrate. Or could anyone on this forum help by showing they would use the algorithm to perform the first two iterations? Many thanks
Below is the dataset I am using. It contains only two rating factors and I would like to perform the algorithm on this dataset. It's a claim frequency model so the distribution is assumed to be Poisson.
Policy years Claims Vehicle Size Vehicle Type
500 42 small 1
1200 37 medium 1
100 1 large 1
400 101 small 2
500 73 medium 2
300 14 large 2
Does anyone know of a resource that might illustrate the approach to the algorithm using a worked example and a couple of iterations to illustrate. Or could anyone on this forum help by showing they would use the algorithm to perform the first two iterations? Many thanks
Iteratively Reweighted Least Squares
