Page 134 - Contributed Paper Session (CPS) - Volume 3

CPS1970 Nurul Aityqah Y. et al.
absolute percentage error (MAPE) and mean absolute scaled error (MASE) to
evaluate the accuracy of the LC model and its variants. The result shows that
BMS method performed well for female, male and total population for the
Malaysian data. Booth et al. (2006) also found that the BMS method have been
found to be more accurate than original LC model and the LM variant. The
decomposition of error has demonstrated that jump-off bias is a significant
source of error for LC (Booth et al., 2005).

Table 2: Evaluation performance of LC method and its variant (Female)
Method ME RMSE MAE MAPE MASE
LC o.82 0.97 0.82 28.28 1.22
LM 0.87 1.06 0.87 27.37 1.23
BMS 0.63 0.81 0.63 19.78 1.06

Table 3: Evaluation performance of LC method and its variant (Male)
Method ME RMSE MAE MAPE MASE
LC 0.28 0.44 0.31 10.88 0.56
LM 0.24 0.44 0.34 10.29 0.57
BMS 0.15 0.31 0.27 7.88 0.51

Table 4: Evaluation performance of LC method and its variant (Total
Population)
Method ME RMSE MAE MAPE MASE
LC 0.44 0.57 0.44 16.27 0.77
LM 0.47 0.65 0.47 15.64 0.76
BMS 0.42 0.58 0.42 13.75 0.78

4. Conclusion
Based on our observation, we can see that the log mortality rates for male,
female and total population decreased as time passed and females in Malaysia
are expected to outlive the male population. The BMS variant performed best
for Malaysian data compared to its two counterparts examined in this study.
In conclusion, all the methods with actual rates taken as jump-off rates
performed better than fitted rates (Zakiyatussariroh et al. 2014).

References
1. Andreozzi, L., Blaconá, M. T., & Arnesi, N. (2011). The Lee Carter Method
for Estimating and Forecasting Mortaity: An Application for Argentina. In
International Symposium on Forecasting-2011, Prague Proceedings,
2050(5), 1–17.

123 | I S I W S C 2 0 1 9