Return to ChangeLog CVS log

Up to [Local Repository] / imach / src

*** empty log message ***

    1: 2003-06-17  Brouard Nicolas  <brouard@homebrou>
    2: 
    3: 
    4: 	* imach.c (Repository): Check when date of death was earlier that
    5: 	current date of interview. It may happen when the death was just
    6: 	prior to the death. In this case, dh was negative and likelihood
    7: 	was wrong (infinity). We still send an "Error" but patch by
    8: 	assuming that the date of death was just one stepm after the
    9: 	interview.
   10: 	(Repository): Because some people have very long ID (first column)
   11: 	we changed int to long in num[] and we added a new lvector for
   12: 	memory allocation. But we also truncated to 8 characters (left
   13: 	truncation)
   14: 
   15: 	(Repository): No more line truncation errors.
   16: 
   17: 2003-06-14  Brouard Nicolas  <brouard@homebrou>
   18: 
   19: 	* imach.c (Repository): Add of a routine likelione (likelihood
   20: 	only once) which prints on a text file
   21: 	(ilk) the contributions to the likelihood for each
   22: 	individual/wave.
   23: 
   24: 2003-06-13  Brouard Nicolas  <brouard@homebrou>
   25: 	* imach.c (Repository): Replace "freqsummary" at a correct
   26: 	place. It differs from routine "prevalence" which may be called
   27: 	many times. Probs is memory consuming and must be used with
   28: 	parcimony.
   29: 
   30: 2003-06-17    <brouard@BROUARD>
   31: 
   32: 	* timeval.h (Module): Added included file to make use of
   33: 	gettimeofday working on win32 with cygwin.
   34: 
   35: 2003-05-16  Brouard Nicolas  <brouard@brouard>
   36: 	* imach.c (Module): Insert of a warning if the delay between two
   37: 	waves is negative.
   38: 
   39: 	* imach.c (Module): 
   40: 
   41: 2003-05-03    <brouard@BROUARD>
   42: 
   43: 	* imach.c: Some cleaning
   44: 
   45: 2003-03-28    <brouard@BROUARD>
   46: 	* imach.c (Module): In version up to 0.92 likelihood was computed
   47: 	as if date of death was unknown. Death was treated as any other
   48: 	health state: the date of the interview describes the actual state
   49: 	and not the date of a change in health state. The former idea was
   50: 	to consider that at each interview the state was recorded
   51: 	(healthy, disable or death) and IMaCh was corrected; but when we
   52: 	introduced the exact date of death then we should have modified
   53: 	the contribution of an exact death to the likelihood. This new
   54: 	contribution is smaller and very dependent of the step unit
   55: 	stepm. It is no more the probability to die between last interview
   56: 	and month of death but the probability to survive from last
   57: 	interview up to one month before death multiplied by the
   58: 	probability to die within a month. Thanks to Chris
   59: 	Jackson for correcting this bug.  Former versions increased
   60: 	mortality artificially. The bad side is that we add another loop
   61: 
   62: 
   63: 	which slows down the processing. The difference can be up to 10%
   64: 
   65: 
   66: 	lower mortality.
   67: 
   68: 
   69: 
   70: 
   71: 
   72: 2002-05-30    <brouard@BROUARD>
   73: 
   74: 
   75: 
   76: 
   77: 
   78: 	* 
   79: 
   80: 
   81: 
   82: 
   83: 
   84: 	* imach.c (Module): Add correlation matrix of one-step
   85: 
   86: 
   87: 	probabilities and covariance matrix
   88: 
   89: 
   90: 
   91: 
   92: