|
imagine if it told you the "per minute" of each action (next to the graph)... i've suggested this before. for example: (yes these are from actual replays)
nada tvz 363 action per minute 179 hotkey per minute 54 move per minute 41 select per minute 28 rally per minute 18 attack move per minute 12 train per minute 5 stop per minute 5 shift select per minute 5 heal per minute
boxer tvt 217 action per minute 72 move per minute 54 hotkey per minute 43 select per minute 12 rally per minute 6 train per minute 5 attack move per minute 5 shift select per minute 3 unload per minute 3 build per minute
|
Russian Federation1020 Posts
Is this just taking the Action Distribution percentage and multiplying by the APM? (ie. .13 * 217) If its some other method, it would be pretty neat, but otherwise, i just think its redundant.
|
Well i'd not find it really helpful since we get the percentages right now anyway, these would just be the real values, not anything new.
|
comparing two player's percentages is different ... this is basically showing you why their apm is what it is, what they are doing. you are thinking about it backwards. with this, you could see, well he's not so fast except for he's doing NNN, or NNN is actually faster in most respects. just looking at percentages you couldnt really say "well boxer moves way more than nada and is almost as fast as nada except for selecting hotkeys"
setting and selecting and double selecting hotkeys dont seem seperated in the actions graph either :/
|
sorry stimow, don't think it'd be useful
|
It wouldn't be useful for me. Knowing my APM isn't even useful for getting better.
|
Vatican City State593 Posts
|
Russian Federation1020 Posts
Unless you could make some graph like the Actions Chart, where it displays the peaks and troughs throughout the game, but only with Hotkeys/Build/Stim/etc. That would be kinda cool, just to see where they use the different actions most, otherwise, the percentages are sufficient me thinks.
|
well, assuming these two reps were the only thing in the world we ever compared, it looks interesting to me. it won't make you a better player but it could help you understand the differences between players and matchups.
for example (looking at the two examples):
nada hits hotkeys over twice as much as boxer did, of course nada's apm is not even close to twice as much as boxer.
boxer moves significantly more often than nada.
boxer hotkeys about as often as nada moves.
nada rallies about twice as much as boxer.
both select about the same amount per minute (boxer insignificantly more).
nada "trains" twice as much as boxer (funny pun i kno).
both shift select at about 5pm or about once every 8 secs.
boxer gives an unload command as often as he gives a build command.
none of that is interesting to any of u? :/ or useful? right now everyone looks at apm and argues about what it means. imagine if you could get into more detail, e.g. "every good tvz has NNNNN per minute and u are XXXX"?? maybe the trends arent seen here but surely there would be some if more reps were looked at in this way, but it can't be done until it's made easier.
|
i think it would be very useful, particulary with seperate action charts.
|
|
|
United States4991 Posts
It wouldn't really be useful, but it'd be cool to know.
|
|
I'd be interested in the update.
|
It potentially could be very useful, nice idea.
|
Cool idea, I think it is useful for figuring out acctual APM. As you can see in this case BoxeR and NaDa are about the same speed even though NaDa has 350 APM to BoxeR's 250.
|
I don't find that idea THAT interesting, but what i know is
I have 350+ replays of myself in which I'm not a newbie anymore, and I'd like to be able to analyze them with bwchart to find out what's the "vs race" distribution for me i.e. what percentage of games is vs toss,zerg,terran.
Then by taking a large replay sample, like 5k or 10k replays, i guess you could find the approximate distribution in all of starcraft population, but that wouldn't work for my 5k replays cause 99% of thhem involve at least 1 terran.
|
well, that was sort of another idea i had, to follow up this idea; mass replays of players could be analyzed and acceptable "profiles" could be created. then it could look at your tvz rep and say, "hey, ur style is most closest to: (player's name here)" which would be good times. there could be "compare my style to player N" etc. of course you could build a profile of any player, the more reps you have of them, the better. u could build a profile of boxer pre-2003 and post-2003 and compare them. all kinds of interesting possibilities arise if these numbers could be combined for multiple reps, groups of reps compared, reps of one player compared to another, reps of one matchup compared to another, any group of reps compared to any other group of reps, any one rep compared to any group of reps, i hope you get the picture. imagine being able to distinctly see something that seperates you from just about every decent tvz player out there? maybe you wouldn't know what the numbers meant, but maybe you would. you could try to find out with experimentation and lasgo's observer pack too.
|
On March 27 2004 17:12 MPXMX wrote:
I have 350+ replays of myself in which I'm not a newbie anymore, and I'd like to be able to analyze them with bwchart to find out what's the "vs race" distribution for me i.e. what percentage of games is vs toss,zerg,terran.
You can already do that if you list the replays by "players".If you used different names, then use the akas to group all your different names into one main name.
|
|
|
|