Ranking System for Ordinal Longevity Risk Factors using Proportional-Odds Logistic Regression

Longevity improvements have traditionally been analysed and extrapolated for future actuarial projections of longevity risk by using a range of statistical methods with different combinations of statistical data types. These meth-ods have shown great performances in explaining the trend movements of the longevity rate. However, actuaries believe that knowing the trend movements is not enough, especially in controlling the impact of the longevity risk. Accessing the effects of each level of the risk factors, especially ordinal risk factors, towards the improvements of the longevity rate would provide significant additional knowledge to the trend movements. Therefore, this study was conducted to determine the potentiality of Proportional-Odds Logistics Regression in ranking the levels of the ordinal risk factors based on their effects on longevity improvements. Based on the results, this method has successfully reordered the levels of each risk factor to be according to their effects in improving longevity rate. Hence, a more meaningful ranking system has been developed based on these new ordered risk factors. This new ranking system will help in improving the ability of any statistical methods in projecting the longevity risk when handling ordinal variables.

@article{Hanafi2020,
  title     = {Ranking System for Ordinal Longevity Risk Factors using Proportional-Odds Logistic Regression},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {11},
  number    = {2},
  year      = {2020},
  publisher = {The Science and Information Organization},
  author    = {Nur Haidar Hanafi and Puteri Nor Ellyza Nohuddin},
  doi       = {10.14569/IJACSA.2020.0110289},
  url       = {https://doi.org/10.14569/IJACSA.2020.0110289}
}