Distance points "feature" in GAP... scoring modules

[Stein Tore Erdal]

17. raised by Yoshiki:

"... question regarding scoring points. Please see the attched file (http://fs.fai.org/raw-attachment/ticket/165/task_result-Jname-stopoutland_2009-03-21_(20090321-2324).html). You can see the distance points of pilot id 8 and 57.

Their distances are same (40.4km) but the distance points are different, 924.3 and 921.4. 2.9 points difference.

Is the reason that the distance is rounded up?"

A "feature" ;-)

The distance score for a pilot comes from two things:

1. a linear factor: distance_points_linear = dist / 2.0 / best_dist

"dist" here is the actual distance the pilot flew (not rounded).

2. a distance difficulty factor: The distance difficulty array is created by looking at the distribution of distances along the course.

A distance difficulty factor is created for each 100m of the task. To find the correct factor for a pilot the distance he flew is rounded down to nearest 100m.

For ex. (in you case),

Avaliable distance points: 982.5

Best distance flown was 42.937

Distance difficulty array:
40.3km: 0.467836257309941
40.4km: 0,470175438596491


Pilot 57: flew 40.358km 
-> 982.5 * (40.358/2/42.937 + 0.467836257309941) = 921.4

Pilot 15: flew 40.390km 
-> 982.5 * (40.390/2/42.937 + 0.467836257309941) = 921.8

Pilot 8: flew 40.413km 
-> 982.5 * (40.413/2/42.937 + 0.470175438596491) = 924.3

Pilot 57 and 15 both end up with the same difficulty factor but get differenct linear factor, hence different distance score.

All three pilots is shown in the result-list with 40.4km since the result stylesheet (task_result.tkr.xsl) rounds the distance to nearest 100m.

It might be more correct to round the distance down the nearest 100m before calculating the linear factor.

Any opinions?

Stein-Tore

[lopezrafa@…]

I met several times with this situation.
This would look better:
Show in the ratings gap with 2 decimal places, rounding set 10 meters instead of 100 meters and do so before the calculation, so if 2 pilots get the same distance get the same score.
It is very difficult to explain skorekeeper pilots when they see that the classification with equal distances obtained different scores.

It's just an opinion

[Eduardo]

I think you should round first, and then calculate points. Same rounded distance, same distance points

Replying to lopezrafa@…:

I met several times with this situation.
This would look better:
Show in the ratings gap with 2 decimal places, rounding set 10 meters instead of 100 meters and do so before the calculation, so if 2 pilots get the same distance get the same score.
It is very difficult to explain skorekeeper pilots when they see that the classification with equal distances obtained different scores.

It's just an opinion

[lopezrafa@…]

Sorry but I write in Spanish and translate with google.
if ud. Rounding can be 100 meters in this situation:
Pilot "A" 54.501 km => 55.6 km
Pilot "B" 54.499 km => 54.5 km

Both pilots landed at the same place
If you rounded up to 10 meters
Pilot "A" 54.501 km => 55.50 km
Pilot "B" 54.499 km => 54.50 km


The pilots should have landed very near the same distance points, 2 to 20 meters can be accuracy of GPS and the drivers know, but 100 meters much so that a pilot has the same points.
Another option is that FS continues round after the calculation in the classification but show 2 decimal places ..

So in case you sample, see attachment.

[lopezrafa@…]

If you rounded 100mts

Pilot "A" 54.549 km => 54.5 km
Pilot "B" 54.451 km => 54.5 km

If you rounded 10 meters

Pilot "A" 54.549 km => 54.55 km
Pilot "B" 54.451 km => 54.45 km

sorry, too bad math ;-)

[Stein Tore Erdal]

Thanks for input :-)

As mentioned earlier "A distance difficulty factor is created for each 100m of the task."

What I've done now is two things:

1. All GAP formulas (2000, 2002, 2007, 2007A, OzGAP...) is changed so they now interpolate this factor when finding the actual distance difficulty factor for a pilot's distance.

2. the task result template (task_result.tkr.xsl) is changed to show two decimals for distance.

These changes will be in the download for 1.2.13.

Stein-Tore

[Stein Tore Erdal]

See http://fs.fai.org/ticket/180

Means the GAP formulas (2000, 2002, 2007, 2007A, OzGAP...) stays as is in 1.2.12.

GAP2008 will get this improvement.

Stein-Tore