Wednesday, 18th July 2018
Puzzles Solved Yesterday: 128
Home | Register | Login | Current Puzzle | Archives | Leaderboard | Forum | Tutorial | FAQ
Forum Index
 
Page 1 of 212>
Wouldn't daily bests be interesting?
ostergaard
Kwon-Tom Addict
Puzzles: 210
Best Total: 48m 12s
Posted - 2008.05.01 11:24:14
I'm pretty new here so this might have been discussed before, but I'm wondering if it wouldn't be interesting if your personal bests of each day were saved in the same way that the personal weekly best is saved? Then it would be possible to compare how good you are at doing seven fast solves in a row compared to the sum of the fastest solve you have for each day. It would give a some kind of a measure on good you are at performing at a steady level. Which I for example have trouble with.
gypsy
Kwon-Tom Obsessive
Puzzles: 3496
Best Total: 29m 51s
Posted - 2008.05.02 01:05:34
....or even further to that, how about an average for each day. Then you could see if you're improving or not. I like to go back over the archives every now and then to see if I can better my time from a year or so ago, and therefore get an idea of whether or not I have improved over time.... Just a thought is all....

Cheers....
Last edited by gypsy - 2008.05.02 01:06:00
MondSemmel
Kwon-Tom Obsessive
Puzzles: 4070
Best Total: 7m 47s
Posted - 2008.05.03 00:54:11
Quote:
Originally Posted by gypsy
....or even further to that, how about an average for each day. Then you could see if you're improving or not. I like to go back over the archives every now and then to see if I can better my time from a year or so ago, and therefore get an idea of whether or not I have improved over time.... Just a thought is all....

Cheers....

Averages of any kind have the problem of those "7d Xh X..." times of frustrating daily puzzles/early on when you just started on this website or something like that. They really, really, really inflate those values. If it was an average over e.g. the past three months and nothing else, it might work, but otherwise it's not really that much of a meaningful number (although it would still decrease if you improved, albeit very slowly).
gypsy
Kwon-Tom Obsessive
Puzzles: 3496
Best Total: 29m 51s
Posted - 2008.05.03 01:49:35
...but, if for example, you have an average of 8h for a Monday puzzle because of those 2d times when you first started, and you are now getting 10 minute times, you can go into the archives and re-visit thos earlier puzzles and re-do them and see how much you have improved. Over time you will notice little things and develop your own way of doing things and recognising patterns that you never did when you first started. By re-doing them, you apply your new knowledge to an old puzzle and get a good idea of how much you have improved from when you first looked at that puzzle.... if you know what I mean....

Cheers....
ostergaard
Kwon-Tom Addict
Puzzles: 210
Best Total: 48m 12s
Posted - 2008.05.03 14:50:57
I think I must agree with MondSemmel. It would be nice to have an average over the last three months (or some 10-12 whole weeks). Then the average would after a while go down if you continuously solve puzzles better and better.

But I'm also still more interested in having the daily personal bests. Then there is still something to fight for every day even if you fail completely on one or two puzzles (like last weekends for me...) and lose some inspiration and don't see that a minute more or less makes a difference for the total week for a while. At least for me who's not as good as most of you are.
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2008.05.03 15:39:57
I agree with you, Ostergaard, though

Quote:
Originally Posted by foilman
I could but I'm trying to steer clear of putting too many "best ever" stats on here... the whole aim was supposed to be "what's happened in the last 7 days" and nothing else, to keep things dynamic. Of course now we have a best-ever champion time and times for all old puzzles solved, so I've already deviated from that!

Foilman also mentioned once (if I recall it correctly) that statistics might encourage cheating.

Quote:
Originally Posted by ostergaard
It would give a some kind of a measure on good you are at performing at a steady level.
Of course you can keep records of best daily times and ranks yourself (like I do). My aggregated daily bests/best week ratio is currently exactly 2 over 3, which makes me a pretty inconsistent player too I believe.
ostergaard
Kwon-Tom Addict
Puzzles: 210
Best Total: 48m 12s
Posted - 2008.05.04 11:42:27
Quote:
Originally Posted by chairman
Of course you can keep records of best daily times and ranks yourself (like I do). My aggregated daily bests/best week ratio is currently exactly 2 over 3, which makes me a pretty inconsistent player too I believe.

Yeah, I could. But I'm to lazy for that.
bear
Kwon-Tom Obsessive
Puzzles: 1066
Best Total: 34m 35s
Posted - 2008.05.06 16:49:43
I think there would have to be a mulligan option to make averages meaningful.  About once a month I'll have a day that's a few hours since my wife wanted to go grocery shopping or my dog started whining to go for a walk.  Then you'd need some way to prevent people from calling a mulligan just because they messed up.  Too tricky in my view.
puzzlescot
Kwon-Tom Obsessive
Puzzles: 1919
Best Total: 23m 14s
Posted - 2008.05.19 08:37:25
Here's my tuppence worth.

Averages won't work, period - this needs to be kept simple.

I, and I assume others, keep an xls so I can see my best time for any weekday, best rolling 7 day time (esp. useful for doing archive puzzles, to see wat I 'would' have got), and 'best of the best' sum of my best individual weekday scores.

That way, if I completely screw up a puzzle, I don't have to wait a week to see if I can be competitive again - there's always a target to beat. Remember 90+% of the people here are only ever competing against themselves.

So, I support the proposal to have 'my best <m/t/w..>day time' against  each puzzle. A best 'sum of these' would be nice, and fairly easy to do if the above info is being stored anyway.

Cheers
Alan
mikey
Kwon-Tom Obsessive
Puzzles: 4098
Best Total: 28m 22s
Posted - 2008.05.19 09:32:14
The mean would be a rather poor measure of average for most puzzlers, but a more robust measure such as the median (or a trimmed-mean) would probably work fine.
ostergaard
Kwon-Tom Addict
Puzzles: 210
Best Total: 48m 12s
Posted - 2008.05.20 15:15:50
Quote:
Originally Posted by puzzlescot
So, I support the proposal to have 'my best <m/t/w..>day time' against  each puzzle. A best 'sum of these' would be nice, and fairly easy to do if the above info is being stored anyway.

This is basically the main idea that I was proposing. I probably didn't word it as good though, hence the discussion that followed was mainly on averages.
foilman
Kwon-Tom Admin
Puzzles: 1721
Best Total: 24m 8s
Posted - 2008.05.22 10:38:46
As of now, your fastest time for each day of the week is being recorded (for puzzles in the current week only, and only on the first solve).
Last edited by foilman - 2008.05.22 14:33:20
foilman
Kwon-Tom Admin
Puzzles: 1721
Best Total: 24m 8s
Posted - 2008.05.22 13:08:04
And now there's a "My Statistics" link from your home page which shows, in addition to the previously stored information on number of daily puzzles solved and your best ever weekly time, your best time for each day of the week (starting from this week) and the total of those best times - which should in theory be the same or lower than your best weekly time but as the stats have only just begun being recorded almost certainly won't be...
ostergaard
Kwon-Tom Addict
Puzzles: 210
Best Total: 48m 12s
Posted - 2008.05.22 14:37:32
Great!
foilman
Kwon-Tom Admin
Puzzles: 1721
Best Total: 24m 8s
Posted - 2008.05.22 14:37:47
I've also added the total amount of time spent solving the weekly puzzles (not archived ones, note) and a simple average solving time.

Similarly for wrongs, though that calculation doesn't take into account the "number of changes" result or final score.
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2008.05.22 15:32:32
Quote:
Originally Posted by foilman
...  [it shows] ... your best time for each day of the week (starting from this week) and the total of those best times - which should in theory be the same or lower than your best weekly time but as the stats have only just begun being recorded almost certainly won't be...

Thanks again very much foilman!

Actually, my best weekly time does coincide with the total of best times since I just joined club 19 
foilman
Kwon-Tom Admin
Puzzles: 1721
Best Total: 24m 8s
Posted - 2008.05.22 15:38:12
Well done!!!
MondSemmel
Kwon-Tom Obsessive
Puzzles: 4070
Best Total: 7m 47s
Posted - 2008.05.22 16:31:12
Thanks for the statistics, foilman! Seeing my personal best times for each day in one spot is really practical! I might be compelled to do even more puzzles now :p.

I don't really like the average time per puzzle though - because the values for the weekly puzzles I did in the past cannot be changed anymore, so whenever I threw a whole week away after making one too many mistakes in a single puzzle (or couldn't finish a puzzle for any other reason), these single puzzles are recorded as me having finished (?) them in up to 7 days (I think these are the problem, anyway). In any case, as I cannot change these values anymore because solving archived puzzles doesn't affect this number, my average time per puzzle is now ~1:46h. I know this number will go down with time, but it will take an eternity and it will never become a meaningful number. If I could just do the archived puzzles (those that are recorded with solving times of several days) again, I could get that number to be more representative, e.g., say, 10-15 minutes( after all, I was much slower when I began solving puzzles here, although I have no idea whether these values are even close) instead of a value that is so obviously wrong.

If possible, I suggest resetting this number, too (e.g. collecting only data from this point forward for the averages) or allowing redoing the archived puzzles to affect this score. Removing unfinished (?) puzzles from the data would drastically improve it, too, although there are also many cases where I finished a puzzle after leaving it alone for a few days (hence still having a solving time of several hours - if unfinished puzzles don't figure into this solving time, then those that I solved after leaving them for several days are the problem)).
Giving everyone the option to reset their averages (or adding a new line called "Average Time Per Puzzle since May 22") would be a great solution, too.

For those curious, these are my statistics as of May 22:

Number Of Daily Puzzles Solved: 371 (although these are only the dailies, I've solved tons and tons of BotMs and done almost all the archived puzzles at least once)
Total Time Spent Solving Them: 27d 6h 54m 24s
Average Time Per Puzzle: 1h 45m 54s

The second and third row don't really represent usable information right now. It'd be great if they somehow could, too!
foilman
Kwon-Tom Admin
Puzzles: 1721
Best Total: 24m 8s
Posted - 2008.05.23 09:37:06
I don't really like the average time thing either - the total time has been recorded for years now but never actually used for anything. I put it in for interest only, really. Pretty much everyone starts off with slow solving times and improves, and it can often happen that you begin a puzzle but get interrupted and don't return to it for a day or two... so the average doesn't really mean too much. Hence why it doesn't appear on any kind of leaderboard!

I guess a reasonable alternative might be the average of all solved puzzles in the archives? Of course it's still not something you can really use to rank people as there's nothing to stop people re-doing puzzles over and over to get fast times on them. But as only you can see your average, you'd only be cheating yourself...?
tobiwan
Kwon-Tom Obsessive
Puzzles: 3951
Best Total: 17m 44s
Posted - 2008.05.23 11:30:17
Just out of interest, I think I've solved all the puzzles in the archives, but what's the total number of puzzles archived?  I've done 1094.
Last edited by tobiwan - 2008.05.23 11:30:27
Page 1 of 212>

Forum Index