Annotation of imach/src/ChangeLog, revision 1.3
1.3 ! brouard 1: 2003-05-16 Brouard Nicolas <brouard@brouard>
! 2:
! 3: * imach.c (Module):
! 4:
1.2 brouard 5: 2003-05-03 <brouard@BROUARD>
6:
7: * imach.c: Some cleaning
1.1 brouard 8: 2003-03-28 <brouard@BROUARD>
9:
10: * imach.c (Module): In version up to 0.92 likelihood was computed
11: as if date of death was unknown. Death was treated as any other
12: health state: the date of the interview describes the actual state
13: and not the date of a change in health state. The former idea was
14: to consider that at each interview the state was recorded
15: (healthy, disable or death) and IMaCh was corrected; but when we
16: introduced the exact date of death then we should have modified
17: the contribution of an exact death to the likelihood. This new
18: contribution is smaller and very dependent of the step unit
19: stepm. It is no more the probability to die between last interview
20: and month of death but the probability to survive from last
21: interview up to one month before death multiplied by the
22: probability to die within a month. Thanks to Chris
23: Jackson for correcting this bug. Former versions increased
24: mortality artificially. The bad side is that we add another loop
25: which slows down the processing. The difference can be up to 10%
26: lower mortality.
27:
28: 2002-05-30 <brouard@BROUARD>
29:
30: *
31:
32: * imach.c (Module): Add correlation matrix of one-step
33: probabilities and covariance matrix
34:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>