Modelling mortality with fuzzy-random lee-carter model

This thesis studies a fuzzy-random extension of the Lee-Carter model, which is one of the popular methods for modelling mortality rates. Age-related variables are fuzzified by using triangular fuzzy numbers and for this purpose a fuzzy regression model is developed and solved. The time-dependent variable is modelled as an ARIMA time series. After obtaining the fuzzy mortality rates, the associated probabilities of death and life expectancy were obtained. In the application part, the central mortality rates of the U.S. male population between 1970 and 2000 were used to predict the mortality rates for the next ten years. Finally, the obtained fuzzy results are defuzzified and compared with the Lee Carter model. In the comparison, it was concluded that the performance of the defuzzified central death rates was better.