Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 42 43 [44] 45 46 ... 222

Author Topic: Dwarf Therapist (Maintained Branch) v.37.0 | DF 42.06  (Read 999718 times)

splinterz

  • Bay Watcher
    • View Profile
    • Dwarf Therapist Branch
Re: Dwarf Therapist (Maintained Branch) v.21.6
« Reply #645 on: May 22, 2014, 11:14:16 am »

hi splinters !
im compil version 21.7, animal age not correct:
http://shot.qip.ru/00knEU-5tfn6Twws/
for dwarf it correct:
http://shot.qip.ru/00knEU-5tfn6Twwt/
i can't reproduce that, my testing has the correct adult/child/babies counts. are you sure you've updated the memory layout files?

Edit: just saw your edit. let me know if it seems broken again..

Skin36

  • Bay Watcher
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.6
« Reply #646 on: May 22, 2014, 11:38:46 am »

u can start new game, for test, and see animals age ?
Logged

splinterz

  • Bay Watcher
    • View Profile
    • Dwarf Therapist Branch
Re: Dwarf Therapist (Maintained Branch) v.21.6
« Reply #647 on: May 22, 2014, 03:28:22 pm »

u can start new game, for test, and see animals age ?
i just fired up a new game and the animal ages seem ok.

thistleknot

  • Bay Watcher
  • Escaped Normalized Spreadsheet Berserker
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.6
« Reply #648 on: May 22, 2014, 07:03:49 pm »

where's the mwdf release?  I dl'd a copy of the latest patched build and it's on 21.2

update:

nevermind.  found it in another MWDFT thread

http://www.bay12forums.com/smf/index.php?topic=132010.msg5218106#msg5218106

I think the version should be sourced on your first page.

http://dffd.wimbli.com/file.php?id=8537

and...

I think the owner of the dt thread for mw should be sourcing it or ceding it to yourself and be hosted on your first thread
« Last Edit: May 22, 2014, 07:15:47 pm by thistleknot »
Logged

splinterz

  • Bay Watcher
    • View Profile
    • Dwarf Therapist Branch
Re: Dwarf Therapist (Maintained Branch) v.21.6
« Reply #649 on: May 23, 2014, 11:59:07 am »

where's the mwdf release?  I dl'd a copy of the latest patched build and it's on 21.2

update:

nevermind.  found it in another MWDFT thread

http://www.bay12forums.com/smf/index.php?topic=132010.msg5218106#msg5218106

I think the version should be sourced on your first page.

http://dffd.wimbli.com/file.php?id=8537

and...

I think the owner of the dt thread for mw should be sourcing it or ceding it to yourself and be hosted on your first thread
it's really just two files; the game_data.ini and the default_gridviews.dtg that separates the MW version from the vanilla version. i was really hoping jodgap would keep the mw files updated as new versions of DT were released, but it was getting so far behind i figured it should be done.

Meph

  • Bay Watcher
    • View Profile
    • worldbicyclist
Re: Dwarf Therapist (Maintained Branch) v.21.6
« Reply #650 on: May 23, 2014, 04:37:00 pm »

I saw a script being posted that shows the momentaey, encumbered speed of a dwarf. Maybe thats worth an addition? You could see just how fast/slow your armorusers, crutchwalkers and haulers really are.
Logged
::: ☼Meph Tileset☼☼Map Tileset☼- 32x graphic sets with TWBT :::
::: ☼MASTERWORK DF☼ - A comprehensive mod pack now on Patreon - 250.000+ downloads and counting :::
::: WorldBicyclist.com - Follow my bike tours around the world - 148 countries visited :::

MeMyselfAndI

  • Bay Watcher
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.6
« Reply #651 on: May 29, 2014, 08:33:20 am »

Request regarding the labor optimizer.

(It may already be in, in which case point me in the right direction?)

I don't like the percentage based system for all dwarves. There are a number of labors where it instead makes sense (to me at least) to have exactly n dwarves assigned - siege operating, etc.

Is there a way to do this with DT?
Logged

splinterz

  • Bay Watcher
    • View Profile
    • Dwarf Therapist Branch
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #652 on: May 30, 2014, 09:34:02 am »

new version is finally up on dffd. change summary:

Version 21.8 Patch Notes
  • fixed typos in game_data, added missing value for a thought
  • fixed various item names and added translation support
  • fixed an issue where children/babies still showed in the preferences and thoughts dock when they had been hidden in the options
  • the thoughts dock can be searched by thought descriptions, rather than just by title
  • animals use adult/child/baby as their 'professions' if they're not war/hunt trained
  • the display name for adult animals is now the caste name
  • tweaked animals to always use a child or baby name before going to a generic placeholder name
  • children and babies are now determined by their age relative to their caste's child/baby age specification, instead of relying on the profession name
  • age in months is now always calculated, since it's sometimes used for animal offspring

Request regarding the labor optimizer. (It may already be in, in which case point me in the right direction?)

I don't like the percentage based system for all dwarves. There are a number of labors where it instead makes sense (to me at least) to have exactly n dwarves assigned - siege operating, etc.

Is there a way to do this with DT?
there currently isn't a way to explicitly set a number of workers for a particular job. this was primarily to keep the optimization plans scalable, so for now at least, you'll have to tinker with the ratios, etc. to get it to the number you want.

I saw a script being posted that shows the momentaey, encumbered speed of a dwarf. Maybe thats worth an addition? You could see just how fast/slow your armorusers, crutchwalkers and haulers really are.
yeah i've made a note of this in the past to possibly create a 'speed' column. i just haven't gotten around to testing it out yet.
« Last Edit: May 30, 2014, 09:36:14 am by splinterz »
Logged

soul4hdwn

  • Bay Watcher
  • make due with what you have
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #653 on: June 01, 2014, 01:16:38 pm »

returning after a long... long break,  trying to stay quiet this time.

i can't find out how to sort/group by migration wave.  it used to be a thing.

edit: thank you memyselfani, swore it was in the simplest spot but i was clearly blind.
« Last Edit: June 05, 2014, 05:56:21 am by soul4hdwn »
Logged

MeMyselfAndI

  • Bay Watcher
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #654 on: June 01, 2014, 03:02:22 pm »

returning after a long... long break,  trying to stay quiet this time.

i can't find out how to sort/group by migration wave.  it used to be a thing.
The drop down list under the toolbar. Group by -> Migration wave

Or is that not what you are looking for?
Logged

thistleknot

  • Bay Watcher
  • Escaped Normalized Spreadsheet Berserker
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #655 on: June 07, 2014, 09:15:43 pm »

I spent a lot of time on this draft, please comment friends.

Update:

I noticed that the grid values are not drawn on role ratings, but rather skill ratings (probably calculated as a CDF).  However, I can work with this, towards the end I propose a custom grid view (based on a labor optimization plan), that draws the raw values of a role that is assigned to a labor (via the optimization plan), as described with the various options below (A through D).  That way a role rating % is drawn vs a % based on skill only (for that specific labor)...


One of my biggest gripes with Dwarf Therapist is...

   The way the role %'s are drawn is not based on anything used by the labor optimizer, nor does it allow for comparison with other role's %'s on the gridview.  Currently, one specific labor's % is showed for all dwarfs.  This % is based on a CDF function ran on just one role's values (i.e. the values listed in one column of a gridview are based on just that column's raw role ratings ran through a CDF).  This means the % is related to other values within that role, but are not comparable to other roles.   
   
   When is it important to compare two role %'s against each other?
   
   When you are comparing two similar roles, like swordwarf or macedwarf; or even a role that is based on a skill only.  So you may want a direct comparison; and this is the value that is used in the labor optimizer.
   
   There is a way to solve this, and it draws on the raw value's being transformed to represent a nicer scale, and to preserve the same sorting/ordering behavior of the columns.

It is very important for a player to understand how the values of different roles compare to each other when setting priority values; allowing a player to see the [raw] values that are used for the labor optimizer would allow for this.

I think this is important for several reasons.

   It shows a player how similar roles can compare with each other.  Sometimes (especially when roles are very similar, say based on only experience, or maybe when only one small part of a role is different between two roles, e.g. "mace dwarf" vs "sword dwarf").

   It allows for a (comparable, sorting preserved) scaling of all raw role %'s into a range from 0 to 100.

   It shows a player the proper ordering of how the roles are used during the labor optimizer


   My proposal:  A drop down button that allows a player to change the the way the grid view values are calculated.
   
   Option A.
   Option B.
   Option C.   
   Option D. As it's currently drawn.
   
   A. I would propose doing a CDF function over all raw role ratings% (i.e. one SDEV for all those raw role ratings calculated for all roles).  That way the value that is listed is still in the same order as the current % ratings, but you get a wide range of values from 1 to 100% from all roles, which would still reflect the same order when sorting each role/column.  Some roles will be shown as all really low, or all really high, but it gives a player a sense of the order the labor optimizer is using, and the sorting/ordering doesn't change between current methods vs this proposal.

   B. Then a CDF function on the results of A but over just a specific role. (i.e. for that specific role %'s ran through a CDF.)

   C.  Labor Optimizer values:  I.E. The values of the raw role rating %'s after it's factored by priority should be listed as an option.  I.e. the values used for Labor Optimization (this may require scaling down prioritie's if they allow priorities higher than 1, then the priorities should all be scaled to a range from 0 to 1 (i.e. 0.xx values).
   
   I also think these values should be available to use with scripts. 
   
Additional request:

   I think a color coded scatterplot that a player can drag roles onto to allow a player to see a scatterplot of
   whatever role's he wants to see compared with each other (say his optimization plan), that way he can adjust priorities accordingly.
   
Labor Optimizer:

   The values used by the labor optimizer (i.e. currently raw values * priority), the raw value should be allowed to be based on other values, say the values derived from Option A, B, or C. 
   
   This way a player can see how roles compare with each other % wise, and choose a way to optimize the dwarves (I think it's still common sense to use Option C,
but I can see the case for wanting to swap it with Option A's values, or even Option A weighted against Option B.)

Edit 6-10-14: Ultimately, I went with this decision, as I could use the average raw values of each role as a weight to adjust the ranges against each other and preserve the proper ordering of roles from highest/lowest; but now I can move the ranges closer towards the center, so more crossover occurs during labor optimization.  Crossover is what you want, as many options to choose from.  And by allowing 4+ labors per each dwarf, you should be able to ensure good crossover opportunity

Update:

A few ideas on workarounds

Allow for grid views to be drawn based on role rating's raw %, maybe even the value outputted by the Labor Optimizer (i.e. after priority is applied).

Update:

On second thought.  I failed to realize that roles do not equal labors necessarily.  There may be many roles involved with a labor.  So, if this were to be applied.  The grid view would most likely need to be related to a labor optimization plan.  That way the role's are tied to a labor, and a % can be used for the gridview.  One could target what columns/roles in the grid view are to be part of this calculation, while other's could be separate.


It is important to understand that role ratings and labors are not the same thing.  I may have screwed up that relationship without even knowing it, so if you notice my proposal conflicts with that, let me know; but I don't think I did.
« Last Edit: June 10, 2014, 07:56:14 pm by thistleknot »
Logged

splinterz

  • Bay Watcher
    • View Profile
    • Dwarf Therapist Branch
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #656 on: June 08, 2014, 06:29:43 am »

Quote
the grid values are not drawn on role ratings, but rather skill ratings
the grid values are drawn based on the adjusted role ratings. by adjusted i mean after they've been compared to all other unit's values.

Quote
I also think these values should be available to use with scripts.
this already exists to some extent. when you call d.get_role_rating the second argument is a boolean flag that indicates whether or not you want the raw role rating, or the adjusted value.

because the grid currently draws 0-100 values, drawing the raw role ratings isn't very helpful. it's basically a sea of red because the role ratings are compared to the potential maximum values in everything, so nearly every role rating has a very low rating.

A. this can be done, the biggest issue is that the role ratings become much more polarized. almost no units end up in the 'middle ground' where nothing is drawn for the rating. they're either very well suited for the role, or very badly suited.

B. this would probably be better for the drawing, as you'd get the middle range back, but again, you're not going to have the transparency as to what the raw role ratings are in the optimizer. the downside is that there will be a bit of a performance hit to calculate all ratings a second time.

**as you can see in the linked images, the sorting for the global vs global adjusted remains the same, it's more to have clearer drawings.

C. this is much more difficult to do, because every role for every dwarf has to be recalculated every time you change optimization plans.

for the optimization plan it might be possible to add another column to show something that could give you an idea of the prioritized ratings. for example showing a column that has the average role rating * priority for the role selected. it wouldn't be precise, but it would give you something of an idea of what the priority would look like.

another thing that could be done with all these different role ratings would be to add a right click sort menu to role columns to choose to sort by the raw value or whatever, although i'm not sure how different the ordering would be.

thistleknot

  • Bay Watcher
  • Escaped Normalized Spreadsheet Berserker
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #657 on: June 08, 2014, 07:46:02 am »

TLDR, see update/conclusion

Quote
the grid values are not drawn on role ratings, but rather skill ratings
the grid values are drawn based on the adjusted role ratings. by adjusted i mean after they've been compared to all other unit's values.

mini-update:

you know what, I'm confusing gridview role rating values with gridview labor drawn values as well.  What i'm referencing is what is drawn when seeing a labor and how it fits into the optimization plan.  However, the way squares are drawn for role ratings also still ARE NOT COMPARABLE with each other, and this discussion addresses that (at least for purposes of the labor optimizer).

could you explain like I'm five?  I don't know what you mean by "they've been compared to all other unit's values"?  Am I wrong is believing it's based on a skill only role?  I.e. a role that matches a skill to a specific labor (ex... for the carpenter labor, the value drawn in the grid, is based on a role that contains only the carpentry skill?).  If this is not the case, I don't understand what other values within the unit the square drawn would be based on.  What are these "other [unit's] values"?

Quote
Quote
I also think these values should be available to use with scripts.
this already exists to some extent. when you call d.get_role_rating the second argument is a boolean flag that indicates whether or not you want the raw role rating, or the adjusted value.

oh that's nice.  That means if I wanted, I could script some of this stuff together (maybe) using the script editor.  However, I would have to have a way to allow the grid view to draw just a value that is outputted by the script, and not necessarily ran through a CDF.  However, another issue would be the inability to run a CDF over just a specific subset of data (i.e. say I wanted to CDF run through just 3-5 roles values), instead i'd have to calculate all involved roles each time for a specific role's script I wanted outputted if I did it a script route.

Quote
because the grid currently draws 0-100 values, drawing the raw role ratings isn't very helpful. it's basically a sea of red because the role ratings are compared to the potential maximum values in everything, so nearly every role rating has a very low rating.


I agree, I believe you.  I have seen the raw role ratings drawn.  I've tested it.  It's because dwarf's couldn't ever really reach 100%. 

My proposal instead, to avoid (long calculations of all values displayed), as well as to avoid MOSTLY RED VALUES; is two fold:

Is to allow for just specific roles to be added to the grid view that are flagged for calculation with method A.  Say I have 4 roles only listed in a grid view, Carpenter, Mason, Farming, and Fishing; these 4 have a special flag set on them to make them pool their raw role rating values into one list, the primary key being their raw role rating value.  Say you have 10 dwarf's.  That's 10 dwarf's x 4 role's worth of values, for a list that has 40 entries.  These 40 entries are the values that are ran through a CDF, one sdev for 40 entries. 

This would give you a range from 0 to 100.  The sort/order operations would still be the same for each column. 

That's the beauty of it.  This way not a lot of values are calculated, and only a specific set of columns in a grid view would have this applied.  That way I could use a spacer in a grid view to section of say a "Labor Optimization Order" section of a grid view, and include just the role's in my labor optimizer.

this would allow for a much cleaner "range" of values (the 100% would be the highest raw value of the 40 values [in the example I just provided of 4 roles x 10 dwarfs], and a player can control it.  This means it will not include every role rating in it's CDF calculation (which would be dumb, because role's that are never even used would have an affect on the grid view's drawn with my proposed methods), but only those the player has marked.  It also means it would scale some values at 100% and some at 0%.  Instead of using red to black values, one could use a increasing black square, or just leave the drawing method as is (in relation to this method).

Quote
A. this can be done, the biggest issue is that the role ratings become much more polarized. almost no units end up in the 'middle ground' where nothing is drawn for the rating. they're either very well suited for the role, or very badly suited.

In your provided picture, I see a lot of units that have no squares drawn.  To me, that would tell me that they are in the median range of all the role ratings for these roles.  However, given no context (such as say 30 roles in a labor optimization plan), I wouldn't really care to see a specific role compared to every other role, but only those that I'm visually looking at (i.e. such as in a gridview together).


Which picture are you referencing to A?  Not default (i.e. 1st pic), but the 2nd pic right?  I think I see what your saying.  I like that you took the effort to draw it out.

I can see why using it as a default view would be dumb, but I think it would be useful in planning out an optimization plan.  Look at Spinner.  Say if that were an important role to me, such as Mining is, and I noticed it was all red.  I would compare that to say... Hive Keeper, which I didn't think was an important role (I think providing averages at the top of a role using this drawing method should be displayed, because that would truly help me in setting priorities).  I could set a very high priority for Spinner, and a very low priority for Hive Keeper, if I saw the averages listed at the top, I could "target a new average" % that I would hope to draw (hence, why drawing optimization plan values would be useful as well, OR AT LEAST OPTIMIZATION VALUES ran through a CDF.  That way I could see a range from 0 to 100%, and I could see the order of how everyone would be "optimized".

Quote
B. this would probably be better for the drawing, as you'd get the middle range back, but again, you're not going to have the transparency as to what the raw role ratings are in the optimizer. the downside is that there will be a bit of a performance hit to calculate all ratings a second time.
[/url]

**as you can see in the linked images, the sorting for the global vs global adjusted remains the same, it's more to have clearer drawings.

Yes, you wouldn't get the transparency of using raw values.

I think the middle range would be found if only applying the scale to a small number of roles, that a user could define.

However, you are right about the transparency, so...  Since raw values are so ugly...

I propose maybe a "hover your cursor over said grid square", and see the raw role rating and/or labor optimization output value (i.e. raw * priority).  That way a player can see the highest raw/opt value for a given square (say if sorting), and lowest raw/opt value for a given column, maybe even the average raw listed above the top of the column.  That way a player can compare roles to each other that way.

Quote
C. this is much more difficult to do, because every role for every dwarf has to be recalculated every time you change optimization plans.

First, my note on what roles would be calculated would be set by a player (essentially I propose having this as a gridview flag, that way it's not redrawn based on optimization plan, but instead a player would have to build a grid view, based on an optimization plan. 

Or maybe better yet, the grid view would be drawn based on the selected optimization plan selected.  This idea has a lot of potential.  If one wants the transparency of the optimization plan, I would propose instead of Option A, a CDF of global raw values (i.e. all roles on a screen), do a CDF of all global optimization values.  That way a player can set all priorities to the same value, and see how the order is preserved when running the optimization plan.).

However...

In direct reference to Option C.

Option C wasn't as important, as this is just a way to draw it as I currently understand the way roles are calculated (i.e. most comparable to the way the gridviews are currently drawn), but if one wanted to get a view of how they are currently drawn, given a gridview that is specially flagged to draw them as Option A.  I would think allowing for this Option C (i.e. global redrawn local, or as you listed, pic 3) to draw the grid view would make logical sense, as it would use the values that Option A outputted, so if a switch were needed to do a per role CDF vs a global, I would think it would allow for quicker calculation using Option A's outputted values (to address your calculation concerns).  As you have pointed out, the sorting is the same (per role, i.e. column) as Option A.

I compared it with the default way it's drawn (i.e. last pic and first pic), Dostust for Mining is the only one in the 1st pic that has a diamond (i.e. best fit for position).  On the way it's drawn in the 3rd pic, he has a diamond as well, but so do a lot of other people.  Which I found to be weird.

NOTE:  I LOVE THESE PICS, BUT THE ORDER isn't the same between them.  If you could output something similar, but each pic sorted by Mining, the order would be the same across all pics, and would give a better comparison of how the values are drawn different between them.

Quote
for the optimization plan it might be possible to add another column to show something that could give you an idea of the prioritized ratings.

for example showing a column that has the average role rating * priority for the role selected. it wouldn't be precise, but it would give you something of an idea of what the priority would look like.

another thing that could be done with all these different role ratings would be to add a right click sort menu to role columns to choose to sort by the raw value or whatever, although i'm not sure how different the ordering would be.

oh man, you are a genius.

say average raw rating (per role basis) listed above the role in the gridview?  Or maybe even in the optimization plan itself, a list of the average raw role rating for a role.  That would help a lot, and would save a lot of time.  It wouldn't be as nice or extensive as a grid view layout, with the average role rating listed at the top of an Option A drawn grid view, but it would allow for proper targetting of priorities in a labor optimization plan.

I don't think the ordering would be different.

Update/Conclusion:

I think at the very least.  If I hovered my mouse cursor over a dwarf, I should be able to see a raw role rating value and/or labor optimization value.  Also, Average raw * priority value for a role could be listed at the top of the column and/or next to each labor in the optimization plan as well would be nice.  That at the very least, would give me an idea of how roles compare to each other for purposes of setting priorities.

More extensive approach:

I think a better idea would be to allow for specific roles to be flagged in a grid view (or a gridview is drawn based on a selected optimization plan), and those values are all compared to each other based on raw rating * priority, and then ran through a CDF function.  I.e. 4 roles x 10 dwarf's = 40 values with 1 SDEV.  If he wanted, he could set all priorities to the same value to get an idea of how the raw role ratings would look if ran through a CDF.

This way a player can see how his values in the labor optimization plan affect the ordering of the labor optimizer.

There is a limitation with setting up a grid view that had flags for roles that were all to be fed through a CDF vs using a labor optimization plan to draw a grid view.  That is with when I want to show two roles compared to each other, but necessarily for an optimization plan (therefore no need to link to one).

Say, I could have a military gridview and specify a melee and a ranged dwarf role that would be specifically ran through a CDF and would only compare those two roles with each other.  Again, this would be the raw role rating of a role.  However, if linked with an optimization plan, it would be nice to allow for the raw rating to be multiplied by a priority (if so desired); however, it would have to be linked to a optimization plan.

Added option:

The ability to sort the gridview columns horizontally by the average (raw * priority) of each role. (i.e. shuffle the columns around so the highest average is listed on the left, and the lowest average is listed on the right.  Gives a player a sense of what values are going to be filled/picked 1st when an optimization plan is applied).

Note:

If the grid view is going to be drawn based on the optimization plan selected, I would recommend including the value of the roles that are listed as "None (Use Labor's Skill)"
« Last Edit: June 08, 2014, 11:21:18 am by thistleknot »
Logged

splinterz

  • Bay Watcher
    • View Profile
    • Dwarf Therapist Branch
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #658 on: June 08, 2014, 11:25:32 am »

yes, it sounds like you're confusing how the drawing is done. labor columns use the skill, role columns use the role's rating. as you've said, and as i assumed when answering, the real concern is that it's hard to compare roles to each other.

the 'adjusted rating' just means what's it's doing now. the raw role rating is passed through a cdf for that role.

Quote
My proposal instead, to avoid (long calculations of all values displayed), as well as to avoid MOSTLY RED VALUES; is two fold
i ran some tests and the extra calculations to compare roles to every other role isn't that bad, so i don't think it's too worrisome.

Quote
I can see why using it as a default view would be dumb, but I think it would be useful in planning out an optimization plan...
the averages at the top of role columns might be interesting to add, regardless of what method is used, so i might play around with that. i'm not entirely convinced that displaying the type A grid is that useful to see, especially since immediately after that you state that the data you really want to see is the average. wouldn't putting that into the optimizer and/or gridview column headers work?

Quote
"hover your cursor over said grid square", and see the raw role rating and/or labor...
i can add the raw rating very easily to the tooltip, but again i'm not sure how this is exactly helpful? same goes for the rating * priority. these both seem like they should be within the optimization plan, not jammed into tooltips.

Quote
build a grid view, based on an optimization plan
are you talking about showing columns for each labor in the optimization plan, and then showing the rating*priority in each cell?

Quote
NOTE:  I LOVE THESE PICS, BUT THE ORDER isn't the same between them
the ordering is the same, they're all sorted by mining. you can see the little indicator at the top of the column. the reason in the first view might have a different sorting is that the top dwarves probably have the same rating.

it seems from your final comments about flagging and views and all this is that what you'd really like is to be able to group roles/labors together, and by doing so have the roles compared to the other's in the group.

i guess this could help to give some context to the role comparison? ie. why compare a military role to a craft role? i assume that the grouping idea is to try and make the role ratings more accurate?

if this is the case, then it was actually proposed by ResMar quite a while back with his idea of 'super labors' (grouping labors into a single labor to be toggled, etc. and by doing so the role ratings would have to be combined).

now if this is what you're talking about, then i think the optimization changes proposed (priority * raw avg column) would probably be all you'd need. rather than having a bunch of single labors that are grouped, new gridviews, etc, you could just use one of these superlabors instead.

« Last Edit: June 08, 2014, 11:28:55 am by splinterz »
Logged

thistleknot

  • Bay Watcher
  • Escaped Normalized Spreadsheet Berserker
    • View Profile
Re: Dwarf Therapist (Maintained Branch) v.21.8
« Reply #659 on: June 08, 2014, 11:36:17 am »

On tooltip raw ratings: An average doesn't address the other part of my goal which was to see the RANGE of values that are used in the labor optimization.  It would be useful to know if my miner's cap out at say 0 to 50% raw (assume I set all my priorities to 1 for this mental excercise), but I find that my raw ratings for hive keeping tend to be 0 to 60%.  If I had averages, I wouldn't see where values cap out, I might want to take into consideration that hive keeping has a ceiling that is 20% higher than my miner's highest rated score (and this harks back to the concerns over some roles which include preferences suppress role ratings because a dwarf will never have all a role's preferences, but he could have max skill, and max attributes, etc).  Having the raw rating in the optimization plan, doesn't allow a side by side comparison of two roles (say armorer and weaponsmith, or even melee dwarf vs ranged dwarf, or swordsdwarf vs macedwarf).

mini-update on range:

Range of raw ratings I think is important.  You understand that a lot of roles will display low when the raw role rating is drawn.  I don't recommend we draw it.  Instead, I propose when a player wants to compare two or more roles directly with each other, being able to group them on a gridview is a good way to do it.  A player can add any number of roles, and see the range of values that each role falls into between the multiple roles.  (Say a range of 0 to 50% for one and 0 to 75% and 0 to 80% will scale up so 80% = 100%, and I can see how much I need to boost the others to raise their cieling to match the others [if that's what I desired]).

This all get's back to my main point.  How I can't compare two roles to each other.

Here's my example

http://imgur.com/evepZrS

It's a custom gridview from the MW Therapist layout.  I modified it to include a role called Melee Dwarf and Ranged Dwarf.

If the feature I'm requesting were implemented.  Instead of a column that had what you see for Melee Dwarf and Ranged Dwarf, you would see a new range of values that would show the two roles and how they compare to each other.  For this example, it is very useful, because they often correlate similar values; however, I have no visual way to compare the two, instead I would have to go into the optimizer to view it.  However, my optimizer does take the desired output I wish to be visualized, and applies it's labor assignment.  That's where the disconnect is.  I can't see how they compare against each other.  If I could see a CDF of their raw ratings next to each other in a special grid view arrangement, I could understand how an optimization plan would apply to them and ultimately decide their worth against each other when assigning assignments.

It's important because if some roles are suppressed due to never fully attainable categories.  I need to see that visually by noticing one role has a lot more upper end dwarf's than the other.  I can adjust things to take that into consideration.  Without a visual representation.  It get's kind of tricky, but an average of outputted results in the optimization plan would be a great step forward.

Comparing two roles to each other by looking at an avg at the top of the column is useful, but it doesn't tell me how well my melee dwarf compares with another role (I think for useful comparisons, important: NOT EVERY ROLE, but only those specified should be included in the calculation, hence flags), say ranged on a one to one basis (and your right, comparing with hive keeping is silly, UNLESS it was part of an optimization plan.  Then I can see how the labor optimizer is going to order each dwarf's rating for each role involved in the optimization).

However, I think ordinal ranking is really what's important.  Ordinal ranking, which would be achieved by whatever columns are flagged to be in a CDF function together, would allow for proper comparison of two similar roles, and how the labor optimizer would select those roles in a optimization run.

That's why I was thinking running just the roles I wanted through a CDF.  Or even better, as you said, raw * priority (i.e. labor optimization output) ran through a CDF, to give me a range.  If I wanted to ever compare raw role ratings of multiple roles via CDF (to get an idea of ordering, similar to resmar's request, but I don't actually propose to merge two roles into one role.  But rather the two role's %'s are compared to each other within the same CDF function).  I could just set the priorities all to 1 if you based this on an optimization plan if I wanted to see what the pre-priority adjusted values are so I could adjust priorities to see how it affects ordinal ranking.

I would also recommend listing the avg's in the labor optimization plan layout, (as well as above the columns?)
« Last Edit: June 08, 2014, 01:11:42 pm by thistleknot »
Logged
Pages: 1 ... 42 43 [44] 45 46 ... 222