送交者: HunHunSheng 于 2005-12-06, 16:13:33:

但可以算某年龄段之间的死亡率，就不是百分之
百了(当然如果年龄层上限是150岁估计也是百分
之百)。

当然算某年龄层的死亡率，要有一定的时间窗口
来算，比如算40-50岁之间的死亡率，应该观察
一定的样本人口，算多少人到50岁没死.多少人
40岁后50岁前死了，多少人被CENSOR了(活到40
岁但未到50岁以后不跟踪了)。

保险公司有种LIFE种TABLE的算法，他们
把被CENSOR的人按照特定比例算部分死
亡，统计学家科学一点，用KAPLAN-MEIER
方法算。

• Asymptotically, we all converge to death...to steal John Maynard Keynes - mangolasi (0 bytes) 2005-12-06, 16:18:27
• “率”is rate, not ratio. - wasguru (18 bytes) 2005-12-06, 16:16:32
  • I guess, is mortality rate the sample version of hazard function? - mangolasi (0 bytes) 2005-12-06, 16:21:04
    • they are really the same thing - wasguru (0 bytes) 2005-12-06, 16:27:16
      • I guess your opinion is that the mortality rate of - mangolasi (289 bytes) 2005-12-06, 16:36:09
        • Oh, but the composition by sub-cohorts of education in cohorts of age - mangolasi (390 bytes) 2005-12-06, 16:41:42
    • HAZARD FUNCTION是关于在NEXT MOMENT死亡的条件概率分布密度 - HunHunSheng (325 bytes) 2005-12-06, 16:26:25
      • But in its sample version (dt and d(people) can not be infinitely small), - mangolasi (227 bytes) 2005-12-06, 16:30:00
        • My recollection of hazard rate is (death)/remaining. - mangolasi (27 bytes) 2005-12-06, 16:37:37
        • surprising, you never learned duration model? - xj (76 bytes) 2005-12-06, 16:36:13
      • 用人的寿命讲hazrd funtion刚开始比较大，然后下降，最后又上升 - HunHunSheng (44 bytes) 2005-12-06, 16:29:38
        • If you are not an actuary, I think you must be a very good student. - mangolasi (96 bytes) 2005-12-06, 16:42:51

加跟贴