English Language & Usage: Multi-Layered Discourse Room

Okay, here's how to optimize the faction for best total score per war. 1. Make a list of all players that have an average positive delta total (so active + passive). 2. Calculate the average passive delta for this group. 3. Kick out all players that are not in group 1 and have a negative passive delta greater than the average of group 1.

Vitaly

Well, it's hard to calculate passive deltas for people who have positive total deltas

Because they are those people who miss very few wars

And so the stats are meaningless (not enough data)

Cerberus

Oh, I thought the second image was all passive?

Vitaly

Yes, it's all passive. I only grabbed the data for wars the player has active_battles == 0 in.

Look at the wars column (number of wars)!

Cerberus

Hmm I see, I thought you had those stats.

Vitaly

1:40 AM

Nah, the server doesn't send that kind of data.

Cerberus

I suppose we'd have to copy the numbers hand by hand over a number of wars on record?

Because all data need to be measured over the same wars, and the wars need to be representative of all wars.

Vitaly

We still wouldn't know which wins were passive and which were active.

Cerberus

Oh, shit.

Right.

Then we would have to use relative deltas.

Although some people's decks will perform better in war A than in war B.

Mr. Shiny and New 安宇

Wow this conversation is completely opaque to me

Cerberus

Hmm.

Hi!

It is about the Game.

Which you ought to start playing by now. Matt has started too, although he is still in denial.

Mr. Shiny and New 安宇

1:44 AM

I'm certain I have no time to play any sort of game these days.

Cerberus

And yet you have time for endless discussions!

Mr. Shiny and New 安宇

Well, duh. That's why.

Cerberus

This game is ideal for work, because 1. it is never real time: you can just walk away without so much as pausing it at any time. 2. It forces you to stop playing after a certain time because you will need to wait for your energy/stamina to refill until you can play again.

@MrShinyandNew安宇 Aww who's that?

Doesn't look like Pasiphaë.

Mr. Shiny and New 安宇

@Cerberus dunno, just something in my RSS feed

Cerberus

@Vitaly What if you made some sort of rough estimate on passive points, based on how much one normally loses in a passive loss, etc.? Because we do have the number of passive wins and passive losses, right?

It would be an approximation, but it shouldn't be that far off. Perhaps 20 % or so (with much higher peaks, of course).

Hmm no, we don't know passive wins and passive losses; we only know "wins", "losses", "passive" and "active", all in numbers of battles.

We'd need the overlap between the categories.

Vitaly

1:50 AM

@Cerberus no, we don't. we have the number of active battles, the number of total wins, and the number of total losses.

Cerberus

Yeah, I just realized when I looked at the column again.

Vitaly

> # 1 - userid, 2 - war_id, 3 - wins, 4 - losses, 5 - points, 6 - points_against, 7 - battles_fought

that's the data I grab for every war

Cerberus

7 means active battles?

Vitaly

yeah

Cerberus

I'm sure some statistical analysis would be possible, but it would be quite complicated.

Post it on statistics.se?

Vitaly

1:52 AM

Hahaha.

Cerberus

For example, what if we calculate the proportion of active battles : score delta?

Vitaly

Eh?

Cerberus

That should bring us somewhat closer to an impression of how the number of active battles influences score delta.

Vitaly

it still varies greatly (some surge, some don't, some have good decks for the current meta, some don't)

Cerberus

So we could extrapolate to "active battles = 0" for those who never have 0.

That is, supposing that we find some regularity.

Vitaly

1:58 AM

argh, i hate central heating

Cerberus

Can I haz these data of yours?

Transcript for

English Language & Usage: Multi-Layer…