|
Selecting coefficient ranges from final
calibration
This is an example of one decision algorithm for selecting
a coefficient set from a final calibration
of Demo_city using the demo200 input images.
Using file control_stats.log:
- Sort file in descending order
using the Lee Sallee
metric
- For the top score: select
the corresponding coefficient values
It is possible for more than one run to have the same score, creating
a tie. In this case select the lowest value.
- For each coefficient: in the
scenario file to be used to derive
forecasting coefficients, using the best performing coefficient
value, set the _START and _STOP values.
- For each coefficient: set the
_STEP value to one (1).
Top 3 scores from demo200 final calibration, sorting
only on the Lee Sallee metric:
|
|
|
|
|
|
|
sort
value
|
|
|
|
|
|
|
initial
coefficient values
|
| Run |
Product |
Compare |
Pop |
Edges |
Clusters |
Cluster
Size |
Leesalee |
Slope |
%Urban |
Xmean |
Ymean |
Rad |
Fmatch |
Diff |
Brd |
Sprd |
Slp |
RG |
| 696 |
0.09178 |
0.70813 |
0.9958 |
0.99999 |
0.62293 |
0.92098 |
0.50055 |
0.965 |
0.9846 |
0.65794 |
0.99831 |
0.99859 |
0.7273 |
1 |
3 |
11 |
87 |
1 |
| 697 |
0.09178 |
0.70813 |
0.9958 |
0.99999 |
0.62293 |
0.92098 |
0.50055 |
0.965 |
0.9846 |
0.65794 |
0.99831 |
0.99859 |
0.7273 |
1 |
3 |
11 |
87 |
10 |
| 698 |
0.09178 |
0.70813 |
0.9958 |
0.99999 |
0.62293 |
0.92098 |
0.50055 |
0.965 |
0.9846 |
0.65794 |
0.99831 |
0.99859 |
0.7273 |
1 |
3 |
11 |
87 |
20 |
| 435 |
0.11017 |
0.69032 |
0.9969 |
0.99991 |
0.62174 |
0.95949 |
0.49951 |
0.91687 |
0.98605 |
0.89465 |
0.91054 |
0.99921 |
0.7301 |
1 |
2 |
11 |
81 |
30 |
| 933 |
0.10363 |
0.77654 |
0.99716 |
1 |
0.5881 |
0.85638 |
0.49902 |
0.95635 |
0.98816 |
0.78703 |
0.98132 |
0.99911 |
0.73023 |
1 |
4 |
11 |
78 |
30 |
Coefficient values used to derive
forecasting coefficients:
| coefficient type |
{_START - _STOP, _STEP} |
| dispersion |
{1 - 1, 1} |
| breed |
{3 - 3, 1} |
| spread |
{11 - 11, 1} |
| slope |
{87 - 87, 1} |
| road gravity |
{1 - 1, 1} |
|