[postgis-devel] test output

[strk]

Ok, I've modified test output.
The error gives the difference between estimated and real
data divided by total number of rows in the table.
I think now everything is more clear.

These tables all contain 8102 rows.
gainAOI still seems to make a difference here.

[lo:test_outputs]$ grep avg strk-points-stats10-gain1
    1   (best/worst/avg)        0       0       +-0
    2   (best/worst/avg)        -0.01   3.33    +-1.74
    4   (best/worst/avg)        0.01    4.09    +-1
    8   (best/worst/avg)        0.01    5.81    +-0.68
    16  (best/worst/avg)        0       5.68    +-0.37
    32  (best/worst/avg)        0       5.92    +-0.22

[lo:test_outputs]$ grep avg strk-points-stats10-AOI
    1   (best/worst/avg)        0       0       +-0
    2   (best/worst/avg)        0.11    -0.71   +-0.36
    4   (best/worst/avg)        0       -0.71   +-0.2
    8   (best/worst/avg)        0       0.69    +-0.09
    16  (best/worst/avg)        0       -0.48   +-0.04
    32  (best/worst/avg)        0       0.4     +-0.02

[lo:test_outputs]$ grep avg strk-multipolygon-stats10-gain1
    1   (best/worst/avg)        0       0       +-0
    2   (best/worst/avg)        0.3     1.27    +-0.89
    4   (best/worst/avg)        0       3.01    +-0.57
    8   (best/worst/avg)        0       3.36    +-0.46
    16  (best/worst/avg)        0       3.57    +-0.25
    32  (best/worst/avg)        0       3.19    +-0.16

[lo:test_outputs]$ grep avg strk-multipolygon-stats10-AOI
    1   (best/worst/avg)        0       0       +-0
    2   (best/worst/avg)        -0.35   -1.01   +-0.65
    4   (best/worst/avg)        0       -2.49   +-0.35
    8   (best/worst/avg)        0       -2.43   +-0.18
    16  (best/worst/avg)        0       -1.83   +-0.1
    32  (best/worst/avg)        0       -0.95   +-0.05

And this is old PG74 estimator (with full rows histogram)

[lo:test_outputs]$ grep avg strk-multipolygon-PG74
    1   (best/worst/avg)        0       0       +-0
    2   (best/worst/avg)        -0.54   -2.33   +-1.09
    4   (best/worst/avg)        0       -3.67   +-0.49
    8   (best/worst/avg)        0       -1.99   +-0.16
    16  (best/worst/avg)        0       -1.3    +-0.09
    32  (best/worst/avg)        0       -0.88   +-0.05

We are not far uh ?

Mark, I'll wait your latest test outputs before putting
gain=AOI/cell_area back.

--strk;