Fig 1.
Trend of life expectancies in Bangladesh (Data: Matlab HDSS; 1974–2016).
Fig 2.
Smoothed log-mortality rates for Matlab HDSS (1974–2016).
Years are plotted using a rainbow palette so the earlier years are shown in red, followed by orange, yellow, green, blue and indigo with the most recent years plotted in violet.
Fig 3.
Fitted parameters of the FDA model for men of Matlab HDSS (1974–2016).
Fig 4.
Fitted parameters of the FDA model for women of Matlab HDSS (1974–2016).
Table 1.
Variation explained by the fitted FDA models for Matlab HDSS (1974–2016).
Fig 5.
Fitted log-mortality rates from FDA model for Matlab HDSS (1974–2016).
Years are plotted using a rainbow palette as before.
Table 2.
Forecast accuracy of FDA method during hold-out period (2007–2016).
Fig 6.
Forecast of log-mortality rates from FDA models for Matlab HDSS (2017–2060).
Years are plotted using a rainbow palette as before. Observed mortality rates are showed in gray lines for reference.
Table 3.
Observed and forecast of life expectancies from FDA method for Matlab HDSS.
Fig 7.
Difference between forecast of e0 and e1 for Matlab HDSS (2017–2060).
The green line represents null difference after which e0 will be larger than e1. The vertical blue line is drawn to identify the possible timing for crossover.
Fig 8.
Prediction interval of e0 by HU, until 2060 for Matlab HDSS.
The blue area represents the 80% prediction interval and the red lines indicate the 95% prediction interval.
Fig 9.
Prediction interval of e60 by HU, until 2060 for Matlab HDSS.
The blue area represents the 80% prediction interval and the red lines indicate the 95% prediction interval.
Table 4.
Coverage probability deviance and mean width of confidence interval (CI) during out-of-sample period (2007:2016) from FDA method for Matlab HDSS.