[Gramps-devel] What is the criteria for "Probably Alive"? From: Gerald Britton - 2005-11-28 21:01 ```I notice that there is a filter called "Probably Alive". I would like to know how that is determined. If the person has a birthdate, but no death date, does gramps assume that the person is probably alive based on calculated age? If so, what is the upper age limit? Is there a preference/parameter for that, or is it hard-coded. Also, what happens if the person has no birth/death date (but maybe his parents/grandparents do). ```
 Re: [Gramps-devel] What is the criteria for "Probably Alive"? From: Alex Roitman - 2005-11-28 21:15 Attachments: Message as HTML ```Gerald, On Mon, 2005-11-28 at 16:01 -0500, Gerald Britton wrote: > I notice that there is a filter called "Probably Alive". I would like > to know how that is determined. If the person has a birthdate, but no > death date, does gramps assume that the person is probably alive based > on calculated age? If so, what is the upper age limit? Is there a > preference/parameter for that, or is it hard-coded. Also, what > happens if the person has no birth/death date (but maybe his > parents/grandparents do). See the probably_alive function in src/Utils.py to get all the details. Briefly, here's the logic and some comments from the code: =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D def probably_alive(person,db,current_year=3DNone,limit=3D0): """Returns true if the person may be alive. This works by a process of elimination. If we can't find a good reason to believe that someone is dead then we assume they must be alive. """ [snip] # If the recorded death year is before current year then # things are simple. [snip] # Look for Cause Of Death, Burial or Cremation events. # These are fairly good indications that someone's not alive. [snip] # If they were born within 100 years before current year then # assume they are alive (we already know they are not dead). [snip] =20 # Neither birth nor death events are available. Try looking # for descendants that were born more than a lifespan ago. [snip] # If there are descendants that are too old for the person to have # been alive in the current year then they must be dead. [snip] # If there are ancestors that would be too old in the current year # then assume our person must be dead too. [snip] # If we can't find any reason to believe that they are dead we # must assume they are alive. In the code, too_old means age_in_years > 110 For descendants, the minimal generation gap is taken to be 13 years. For ancestors, average generation gap is taken to be 20 years. None of the above values is user-adjustable. You have to tweak the code to modify them. Alex --=20 Alexander Roitman http://www.gramps-project.org ```
 Re: [Gramps-devel] What is the criteria for "Probably Alive"? From: Alex Roitman - 2005-11-29 18:30 Attachments: Message as HTML ```Gerald, On Tue, 2005-11-29 at 09:55 -0500, Gerald Britton wrote: > OK -- I've attached a bziped copy of my dataabase. Just run filter > specifying people probably alive in 2005 and look for Darryl Beckett > (I0835). He is alive, as are his parents, but I do not know their > birth dates. I figured the reason for this, and unfortunately there's no good solution at the moment. As I said, the person is thought of as probably_alive if there's no evidence suggesting the opposite. So your Darryl has no info indicating death. No descendants to indicate that he must be too old to be alive. So the algorithm takes us to his ancestors and evaluates how old they were 20 years ago. Nothing, since his parents have no birth/death data. So we move one generation up and evaluate those guys another 20 years ago (1965). Nothing (grandfather is OK with birth date, grandmother is OK with both birth and death dates). So we move one generation up and evaluate his g-g-grandparents in 1945. And one of those (paternal g-g-grandfather) was too old to be alive in 1945. So we declare Darryl as not probably_alive. The proper thing to do would probably be stop looking back after a couple of generations. The allowed 20 years per generation is very approximate, and when you go N generations up the uncertainty multiplies by N. So every next generation estimate is less precise than previous. The problem is that if we find no death evidence we just keep looking further. I don't know what would be a good place to stop here. Maybe we should simply accept that probably_alive is an approximate estimate and is not definitely_alive :-) Any reasonable suggestions are welcome, Alex --=20 Alexander Roitman http://www.gramps-project.org ```