This is actually mainly for the granite state regional, as the other regionals didn’t put their scores on usfirst.org
Red Average Score: 23.8
Blue Average Score: 25.4
Average winning margin: 18.2
Median winning margin: 11
Average score for a winning team: 33.7
Average score for a losing team: 15.4
Zero-related
% of matches with one team with zero: 35
Average score for a winning team where the losing team didn’t get zero: 47
Average score for a losing team that didn’t get zero: 23.8
Overall average score, excluding zeroes: 35.6
Interpretation: Half of all matches could be decided by 2 robots on the ramp, since the median score difference was 11. Blue does not have a statistically significant advantage over red (trust me on this one). Scoring a single point will win you your game 35% of the time, although as the graph below shows, most of the zero-scores occurred in the first 75 games, and were pretty rare afterwards. So scoring a single point very early in the regional will win much more often than 35%.
I can’t think of any other numbers that people might want to know about. Hopefully as the weeks go on I’ll update this and the graphs will get smoother as more data is collected.
Here are two graphs. The first is a histogram of all the scores. It shows the proportion of scores that fell into each of the ranges indicated on the x-axis. The second is a plot of the scores versus the time the score occurred in the regional. The x-axis indicates double the game number.
However, if you take the average of the team numbers involved in an alliance and plot them against the score that that particular alliance managed in the game, then a correlation becomes more clear. Example: If we had an alliance of team 10, team 20, and team 45, then their average would be 25 and that’s where they would be on the x-axis.
So, it isn’t your individual team’s age that scores high, it’s the COMBINED age that matters. The higher your age (and the lower your average team number), then the higher you may score.
It would be interesting to band the team numbers by age in years. I’m guessing it won’t make much difference, but it might show more than the scattergram. Did you plot a linear regression and calculate the SD?
I’m using a free statistics program called R. It did the linear regression for me. The slope was -0.003130, meaning that there was a very slight trend towards less scores as team number increased. However, I don’t remember enough from my last stats course to determine the variability of the slope.
So I guess saying “there is no statistical correlation” isn’t entirely true, since I didn’t check. There doesn’t appear to be any correlation, but there may be.
For the difference between blue and red scores though, I did do a check and there is genuinely no statistically significant difference between them.
Edit: found a team age list on the FirstWiki. I’ll try and plug that into my numbers.
I would like to see some data on the split between 1 pt., 3 pt., and ramp points. Where are the points coming from in these matches?
Specificly, I want to know what kind of points are being scored in each mode of the match, in each match, and what kinds of ratios are winning matches in qualification and final matches.
Additional things like eliminating points scored by human player ‘hail mary’s’ would be nice, but probably never going to happen.
Does anyone know if this kind of raw data could be saved by the field? Obviously the feild is keeping track of points scored in diffrent goals at least untill the scores are displayed. But what happens after the scored is entered and the field gets reset? What about the ref score cards? Do they keep track of how many points are scored into each goal in case of a faliure of the field or do they just keep track of final scores?
I think this is all really usefull information! Keep it up!
This is one of those times where you can see that the slope is approximately zero, and the visual plot is all over the place, and conclude that there is no meaningful relationship between score and team number. There is no need to get too fancy here.
Those are the rankings, not the actual match data. Hopefully the match data is correct.
Anyway, here’s a new one. This is team age versus scores, and AVERAGE team age versus score. Note the difference. The 1st one takes the combined experience of an alliance and what their score was, while the second one simply looks at how old each individual team was, and how they did. The extent of the boxes goes from the 25th to 75th percentile. The lines above the boxes go to the 95th (I think) percentile. The dots above them are outliers. You can see that age isn’t a very good predictor of performance. But in the first graph, you see that the sum of multiple old teams does make a noticeable difference.
Edit: Keep in mind that there will be a larger number of averages in the middle of the graph, since it isn’t likely to pick mutiple teams that are all very old or very young. Therefore, it may not be that 6-7 average age alliances are better, it may just be that they are more plentiful. A histogram would be able to show if this is what’s happening, but I left my excel file at work and will have to try tomorrow.
Alright, here’s a histogram to show the number of alliances that had each average age. As you can see, 6-7 year old alliances weren’t all that common, meaning that there is actually a correlation between alliance age and score.
