| Article ID: | iaor19991272 |
| Country: | Netherlands |
| Volume: | 98 |
| Issue: | 2 |
| Start Page Number: | 419 |
| End Page Number: | 430 |
| Publication Date: | Apr 1997 |
| Journal: | European Journal of Operational Research |
| Authors: | Kramer Bert |
| Keywords: | risk, statistics: multivariate, artificial intelligence: expert systems, neural networks |
This paper describes a model which can be used by the Dutch insurance supervisor to determine the priority a non-life insurer should have for further examination. This model combines a traditional statistical technique (an ordered logit model) with artificial intelligence techniques (a neural network and an expert system). The output of the model consists of the priority for further examination (high, medium, or low), and a report which summarizes the main findings of the model. The model was able to adequately classify 93% of the companies in a 1993 data set.