RPrentki Challenge: Results are In!
On Monday, EyeWire HQ challenged you to battle with the @RPrentki Challenge. Rachel, AKA @rprentki and the tour de force behind EyeWire’s Omni Room (and summoner of Grim Reaper), invited all EyeWirers to take on her expert mapping skills.
There were two ways to win:
- Outscore Rachel (most points scored receives double points during power hour).
- Beat Rachel’s accuracy for a +5,000 point bonus (how we calculate accuracy).
After factoring in trailblaze bonuses, we’re thrilled to congratulate @jamiexq, winner of the RPrentki Challenge points division. With her glory comes a +9,683 bonus. Talk about a close race..Jamiexq beat @a5hm0r by a mere 20 points! A5hm0r will receive a +1,000 bonus for such a close second. Check out the full points-based leaderboard here.
Congrats to the top 10 by points:
1. jamiexq 9683
2. a5hm0r 9664
3. rprentki 7942
4. reb1618 7570
5. susi 7342
6. couc 6440
7. nkem 4032
8. marika 3814
9. jinbean 3752
10. Laurcifer 2820
Accuracy Ranking
Accuracy was calculated only if a player submitted 15 starburst cubes during the rprentki challenge. Only starburst cubes factor into this metric. It took a little longer than expected but we’re pleased to announce that accuracy results are in!
Seven players beat @rprentki in accuracy and will receive a bonus of +5,000 points: marika, reb1618, nkem, jinbean, a5hm0r, Laurcifer and susi. Congratulations!
Here is a full list of rprentki challenge accuracy stats. Stay tuned for more EyeWire Games — next friday we’ll feature the nkem trivia challenge, details coming soon!
See you online at EyeWire.org!
——–
Refresher: What is F-score and what do the other values in the rprentki stats mean?
(Text below originally appeared in results to Ketta Competition)
How do we determine F-Score?
First we look at volume added by each player. We separate out this number into Positive Volume (“PV” i.e. correct volume), False Negative (“FN,” i.e. volume that should have been added but wasn’t – these are missed branches) and False Positive (“FP,” i.e. volume added that should not have been — this is where mergers come from).
The F-Score is a combination of precision and recall.
Precision is how much volume was accurately added. It is calculated as follows:
| precision = (PV) / (PV + FP) |
Recall is a metric of how much volume was missed. This is particularly important for Starburst neurons, as volume missed tends to lead to missing branches that don’t spawn new cubes. It is calculated as follows:
| recall = (PV) / (PV + FN) |
This leads to F-Score:
| F-Score = 2 * (precision * recall) / (precision + recall) |