[Babase] Re: Agnonisim and grooming errors

Karl O. Pinc kop at meme.com
Wed Mar 14 23:45:36 EDT 2007


On 03/14/2007 07:27:04 PM, Leah Gerber wrote:
> Karl,
> 
> I am at Susan's lab now. I looked through the agonism and grooming
> data.
> Since 1994 grooming data has only been entered assuming the first of
> the
> month. There is other data in babase because it comes from the psion
> and
> has the exact date. So, I guess it needs to be documented that the
> period
> between January 1994 and June 2006 inclusive contain hand entered
> grooming
> data for the first of the month only, keeping in mind that other
> grooming
> data coming from the psion will have exact dates. The agonism data
> seems to
> be consistently entered with the exact census dates.

I don't believe this to be the case:

babase=> select extract(day from date) as day_part
               , count(*)
            from interact
            where act = 'A'
                  and date < '1994-01-01'
           group by day_part
           order by day_part;
   day_part | count
-----------+--------
          1 | 133969
          2 |    362
          3 |     76
          4 |     37
          5 |    313
          6 |     23
          7 |     51
          8 |     18
          9 |     58
         10 |     26
         11 |      4
         12 |    307
         14 |     79
         16 |     16
         17 |    144
         18 |     33
         19 |     12
         20 |      4
         22 |      5
         23 |    134
         24 |     85
         25 |      1
         26 |      8
         28 |      4
         29 |      1
(25 rows)

None of this agonisim data is psion data.
There's _way_ to many agonisims on the first of the month.

If a user sees one of the first-of-the-month agonisims in this
group then what's she supposed to think about the date?  Is
it accurate or is it an artifact of data entry?

The same problem appears with grooming before 1994.

select extract(day from date) as day_part
      , count(*)
  from interact
  where act = 'G'
        and date < '1994-01-01'
  group by day_part
  order by day_part;

   day_part | count
-----------+-------
          1 | 98097
          2 |    82
          3 |    76
          4 |    52
          5 |    50
          6 |    20
          7 |    25
          8 |    15
          9 |    43
         10 |    18
         11 |    24
         12 |    14
         13 |     6
         14 |    20
         15 |     9
         16 |    21
         17 |    46
         18 |    39
         19 |    30
         20 |    23
         21 |    22
         22 |    18
         23 |    10
         24 |    27
         25 |    11
         26 |    10
         27 |     2
         28 |     1
         29 |     4
(29 rows)

I also find it really wierd that neither of the above queries
found any data for the 30th or 31st of any month.

Grooming is all 1st-of-the-month since 94, but what
about agonisim?

select extract(day from date) as day_part
      , count(*)
  from interact
  where act = 'A'
        and date >= '1994-01-01'
        and date < '1996-07-01'
        and sid is NULL
  group by day_part
  order by day_part;

   day_part | count
-----------+-------
          1 |  8488
          2 |   498
          3 |   508
          4 |   280
          5 |   366
          6 |   637
          7 |   463
          8 |   502
          9 |   557
         10 |   490
         11 |   656
         12 |   617
         13 |   488
         14 |   456
         15 |   665
         16 |   802
         17 |   540
         18 |   625
         19 |   857
         20 |   762
         21 |   530
         22 |   510
         23 |   658
         24 |   568
         25 |   565
         26 |   507
         27 |   595
         28 |   185
         29 |   189
         30 |    81
         31 |    16
(31 rows)

Again, there's too many first of the month agonisims.

Did you look at errors_v10 and loadscript_v10 to see the
problems that the conversion program is picking up?

> 
> Whenever we start using PHP for uploading (I am assuming starting with
> the
> July 2006 data) we will enter the exact dates for grooming again.
> Nothing
> will change for agonisms or mounts and consorts. Does this make sense?

Yes.

> You can call me if you want to talk about it

Likewise.

> Leah


Karl <kop at meme.com>
Free Software:  "You don't pay back, you pay forward."
                  -- Robert A. Heinlein



More information about the Babase mailing list