SURVIVAL MODELS AND INTERPOLATION METHODS FOR INDONESIAN MORTALITY TABLE IV

• Main Author: Ramadhani, Fikri
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/76317
• Institution: Institut Teknologi Bandung

Mortality table is used to analyze death patterns in population of a country. In Indonesia, the latest mortality table is Indonesian Mortality Table IV (TMI IV). To construct TMI IV requires mortality data from several life insurance companies. The number of death data that used to calculate the probability of death must be eligible according to specific criteria for credible result. If the number of data is insufficient, then extrapolation is performed. Especially for very young age (0- 15 years old) and very old age (above 79 years old). Extrapolation can be done using survival models such as the Gompertz’s mortality law, Makeham’s mortality law, Heligman-Pollard’s survival model, and Siler’s survival model. For ages 16- 79 years, probability of death in TMI IV are derived from actual data, but there are often missing or unrealistic data. So, the interpolation is utilized to estimate the probability of death. The interpolation method that used is 6-point Lagrange interpolation, Brass-Logit, linear spline, and cubic spline. Survival models and interpolation methods will be tested for prediction accuracy of TMI IV with statistical test, specifically the mean absolute deviation (MAD) test. In TMI IV for males, the best survival model for very young age is Siler survival model and for very old age is Makeham’s mortality law. For females, Siler survival model is the best for very young ages, and the Heligman-Pollard survival model is the best for very old ages. Among the four interpolation methods, 6-point Lagrange interpolation is the best result.