Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 115 116 [117] 118 119 ... 222

Author Topic: Dwarf Therapist (Maintained Branch) v.37.0 | DF 42.06  (Read 999797 times)

Dirst

  • Bay Watcher
  • [EASILY_DISTRA
    • View Profile
Logged
Just got back, updating:
(0.42 & 0.43) The Earth Strikes Back! v2.15 - Pay attention...  It's a mine!  It's-a not yours!
(0.42 & 0.43) Appearance Tweaks v1.03 - Tease those hippies about their pointy ears.
(0.42 & 0.43) Accessibility Utility v1.04 - Console tools to navigate the map

RoboLuddite

  • Escaped Lunatic
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1742 on: October 28, 2014, 05:28:43 pm »

I, too, want to say thanks. I'd resigned myself to playing without Therapist for quite a while after this release so I was pleasantly surprised to see an update. I really appreciate the effort you're putting in!
Logged

Buttery_Mess

  • Bay Watcher
  • 11x11
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1743 on: October 30, 2014, 02:09:39 am »

The 40.14 memory layouts work in Linux :)
Logged
But .... It's so small!
It's not the size of the pick that counts... it's the size of the man with the pick.
Quote from: Toady One
Naturally, we'd like to make life miserable for everybody, randomly, but that'll take some doing.

AriRashkae

  • Bay Watcher
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1744 on: October 30, 2014, 06:05:21 am »

[OSX] Dwarf Therapist 28.pre-release | DF 40.14

Thank you both so much, for creating (splinterz) & porting (fricy) this marvel!
Logged

94dima94

  • Bay Watcher
  • For !!SCIENCE!!
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1745 on: October 31, 2014, 04:23:57 am »

EDIT: Nevermind. Sorry, I tried to report a bug when there was nothing wrong.
« Last Edit: October 31, 2014, 06:31:07 am by 94dima94 »
Logged
The worst enemy you'll ever meet is the little voice that says Hey, this would be a cool/awesome/funny thing to do!/I wonder what happens if...
"Rock needs a nerf! Paper is just fine, IMO." - The Scissors.

thistleknot

  • Bay Watcher
  • Escaped Normalized Spreadsheet Berserker
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1746 on: October 31, 2014, 09:27:20 pm »

Spoiler: How I monte carlo'd DF (click to show/hide)
Variability in Weighted Averaging vs Weighted Factoring and how to address

Near the end of the setup, I noticed that weighted factoring and weighted averaging had a proportional relationship to each other.  However, something else stood out.

A weighted factor seems to be affected moreso by the # of items it is factored against.  This is because it will ultimately be reduced lower than the lowest factor.  This is not the case with weighted averages.  In fact, weighted averages seem to proportionate the loss equally below and above the new value.

This is where the parallel occurs between weighted averages and weighted factors.  The more numbers of elements in the adding or factoring stage, the more the variability is reduced in terms of upper and lower limit (for weighted averages).  It seems [the output of] weighted averages hover around a midpoint, where as weighted products just reduces the max.  With factoring the CEILING is literally lowered by whatever one is factoring against.  However with averaging it is in a less so way.  The factored output can never be higher than the lowest factored.

Point being... the variability is affected by the # of elements.  Mitigation steps have already been put in place to ensure that the [lack of] # of elements specified in a role don't become a factor.  Such as when a role is missing an aspect (such as when a global weight is disabled), what will happen is a .5 value would be substituted for any weighted average values involving a disabled weight.  Without this, a role with only a skill specified would be displayed as a higher value because it didn't have any other variables to hold it down by).

Initially I thought my issue with weighted factor was specific to weighted factors, but then I realized that no, steps had already been taken to mitigate it in the weighted average formula.  So that's where I'm at right now, I have my weighted product demo ready, but their is a correlated relationship between the # of elements specified in a role and the variability in the min/max the role can achieve.  There are a few ways to address this, but I think ultimately, the roles might be biased against each other in variability by the # of elements assigned in them.

I think ultimately to get a "fair" picture of elements between roles... is to "fake" missing elements.  One idea I had was to supply .5 for the missing element.  I was attempting to do a max count of the # of elements each "aspect" in each role had, aspect defined as Attributes, Personality/Traits, Preferences, Skills.

So I'd ultimately end up with
Attributes Max: ex: 5
Preferences Max: ex: 7
Skills Max: ex: 3
Traits Max: ex: 2

That way, instead emulate a .5 for each missing aspect/element... i.e. one could just emulate that # of elements that would be needed to equalize all roles
« Last Edit: October 31, 2014, 09:43:36 pm by thistleknot »
Logged

Icefire2314

  • Bay Watcher
  • Programmer and Space Enthusiast
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1747 on: October 31, 2014, 10:31:39 pm »

How is emotional state representation looking at being addressed? It was simple to put an emoticon down regarding mood, but that's largely been done away with now.
Logged
"ERUTH PULL THE DAMN LEVER THE ZOMBIES ARE ABOUT TO GET INSIDE!"
"zzz"
BAY 12 MINI CITY: http://bay-12.myminicity.com/

splinterz

  • Bay Watcher
    • View Profile
    • Dwarf Therapist Branch
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1748 on: November 01, 2014, 09:13:56 am »

How is emotional state representation looking at being addressed? It was simple to put an emoticon down regarding mood, but that's largely been done away with now.
Quietust has done some work on this, and from what i gleaned it's something like this now: thoughts are tied to emotions which modify a stress level. depending on the personality, a thought can cause different emotions for an individual and change the stress level in different ways.

so it may be possible to convert the new stress level to a general happiness indicator, but i don't know yet for sure.

Putnam

  • Bay Watcher
  • DAT WIZARD
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1749 on: November 01, 2014, 02:27:46 pm »

Not just stress, but stability. You may be able to get an insanity chance of sorts.

katwithk

  • Bay Watcher
  • Urist likes kats for their calmness and logic.
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1750 on: November 01, 2014, 03:57:20 pm »

Any estimates as to when we'll see a v40.14 compatible DT for OSX?
Logged
A tall, slender creature fond of coffee and computers.

katwithk

  • Bay Watcher
  • Urist likes kats for their calmness and logic.
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1752 on: November 01, 2014, 04:49:39 pm »

Mucho aprecio.
Logged
A tall, slender creature fond of coffee and computers.

pisskop

  • Bay Watcher
  • Too old and stubborn to get a new avatar
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1753 on: November 01, 2014, 04:50:36 pm »

ptw.  Finally.  I occasionally forget to check for updates :(


Logged
Pisskop's Reblancing Mod - A C:DDA Mod to make life a little (lot) more brutal!
drealmerz7 - pk was supreme pick for traitor too I think, and because of how it all is and pk is he is just feeding into the trollfucking so well.
PKs DF Mod!

thistleknot

  • Bay Watcher
  • Escaped Normalized Spreadsheet Berserker
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.28.0 | DF 40.14
« Reply #1754 on: November 01, 2014, 07:38:44 pm »

blah blah blah

So... here it is...

http://dffd.wimbli.com/file.php?id=10014

the crazy Monte Carlo simulation of Dwarf Fortress I built up to test out how Weighted Factoring would affect roles.

it's an extensive (re)work on (re)building [a mockup of] how Dwarf Therapist derives roles (inside a spreadsheet workbook).  It's about 98% accurate to 'as is' calculations.  I say 98% because I'm not 'Monte Carlo'ing Preferences, so I have 3/4 aspects... I also don't do anything special with skills and/or attributes and their special weight adjustments [for skill rate and attribute potential weight] that have their values pre-prescribed before they are ran through 'as is' transforms...  I also say 'as is' because I modified one formula "rank-ecdf" but it's a minor adjustment that doesn't affect Weighted Averaging, but is essential for weighted factoring.

I also introduced a proposed change to allow for weighted factoring, but I'm not 100% sure on it's contribution, how I implemented it, and would love input.  Ultimately, weighted factoring produces very small values, so the value isn't relevant to a central mean like weighted averaging does.  Instead I derive the rank ordering of the values and their respective %'s (i.e. rankecdf) and derive a flat distribution curve to average against.  This is the big question, do you guys like the idea or is it garbage.  Hell, I should involve maklak on this one.

File opens in libreoffice

I'm using numbers generated from [15 dwarfs] raws (for attributes and traits), and the interpolated level for skills (which for this case is vanilla skill rates, so nothing too weird with those)



Sheet: Global Weights & Roles Summary

I was proposing combining influence from Weighted Factoring as I understood it in AO11:aq20

however, a more "sane" as-is approach is deriving a s transform, and a rank-ecdf and average them.

as11:au20

I find my weighted factor approach has the affect of identifying the weakest link, as each value is successively factored against another value, the end result is a number somewhere under the lowest value that was in the list to be factored...  That's the issue with correlated pairing, or whatever I should be calling it.
« Last Edit: November 01, 2014, 08:23:30 pm by thistleknot »
Logged
Pages: 1 ... 115 116 [117] 118 119 ... 222