1,031
edits
- Welcome to the Kancolle Wiki!
- If you have any questions regarding site content, account registration, etc., please visit the KanColle Wiki Discord
Changes
Jump to navigation
Jump to search
<tabber>
|-|Methodology=
Just how much more important one stat is than another is hard to determine, especially when considering different map scenarios that could strongly encourage one stat over another.<br/>
However, the overall rating of a ship could be estimated with some (somewhat) basic math.<br/>
Each ship class is separated into their basic classes instead of separately, as most sub-classes perform largely similar roles (such as CA & CAV under the CA umbrella).abilities/traits that differ from the norm of their class.
Every ship is scored based on its stats, special attributes/abilities, and class.<br/>
* These small differences are instead handled by awarding points to each ship based on special
** For example, CAVs and Zara/Pola get bonus points for the ability to carry Seaplane Fighters.
* Each class also might place more or less importance on an individual stat.<br/>
** With this in mind, each stat/ship class combination can receive a scaling modifier via a stretch/compress multiplier to each stat on a class-by-class basis.<br/>
|-|Averages=
Each basic class has its stats averaged across the entire class, using the stats from each ships' final remodel.<br/>
* Special versions of the same ship are factored in separately from each other.
** For example, Kaga Kai Ni and Kaga Kai Ni Go both apply separately to the overall average.
|-|Stretch/Compress=
Each basic class also has a stretch/compress multiplier based on an estimation of how important an individual stat is to that class of ship.
* In most cases, stretch/compress values remain at '1' due to being universally important and/or there is enough variety between individuals to already have meaningful differences
** A value <1 indicates a 'Stretch', meaning the stat is '''less''' valuable
** A value >1 indicates a 'Compression', meaning the stat is '''more''' valuable
* Exceptions:
** Fuel & Ammo
*** No scalar multiplier
*** Instead, they have additive modifiers to the StdDev for each class
** '''Ignored:''' HP, Luck, LuckMax
|-|Standard Deviation=
The Stretch/Compress value is applied (via multiplication) to the Standard Deviation of each respective stat for each respective class.<br/>
This Standard Deviation value denotes how wide a standard 'step' is from the average for the class which directly factors into the rating clusters.<br/>
The following table displays the Averages, Standard Deviations, and Stretch/Compress values for each class.<br/>
{{User:Memetails/Averages Deviation and StretchCompress}}
|-|Stat Evaluation Clusters=
Using the above values, each individual ship is evaluated into clusters based on their stats.
The math & clusters are as follows:
{| class="mw-collapsible mw-collapsed wikitable typography-xl-optout" style="width: -moz-max-content; width: -webkit-max-content; width: max-content; text-align: center; vertical-align: middle; "
|-
! Cluster
! '''DistanceFromAvg''' = ShipStat - ClassAvg
|-
! Insane:
|> 2 * (StdDev * Stretch)
|-
! Outstanding:
|> 1 * (StdDev * Stretch)
|-
! High:
|> 0.5 * (StdDev * Stretch)
|-
! Normal:
|< 0.5 * (StdDev * Stretch) '''&''' > -2 * (StdDev * Stretch)
|-
! Low:
|< -2 * (StdDev * Stretch)
|-
! Bad:
|< -3 * (StdDev * Stretch)
|-
! Terrible:
|< -4 * (StdDev * Stretch)
|-
| colspan="2"; style="text-align: left;"| Note: 'Stretch' refers to both Stretch & Compress
Clustering Exceptions:
* Fuel & Ammo: Additive Stretch/Compress instead of multiplicative, table is basically inverted due to lower = better
* Luck & LuckMax: Set values shared across all classes.
**Insane/Outstanding/High/Normal/Low/Bad/Terrible = > 50/35/25/10/5/0/NA
* HP: Instead based on 'good/bad' values via the overkill mechanic, such as %4, %6, %12, %4-1, etc.
|}
|-|Stat & Rating Multipliers=
Each stat is assigned a numerical multiplier based on an estimation of total value/importance.<br/>
Each cluster is also assigned a numerical multiplier based on an estimate of how good or bad being in that cluster is.<br/>
These 2 values are then multiplied together, resulting in the following table:
{{User:Memetails/Stat & Rating Multipliers}}
For example:
* Firepower is generally more important than LoS: '''Firepower > LoS'''
* Night power is generally more important than Firepower: '''Night > Firepower'''
* It can be concluded then that: '''Night > Firepower > LoS'''
* The 'Normal' cluster is left out of the table as it has a value of 0, being neither good nor bad
Using this table, each ship can then add together the value for each stat/cluster combination it's part of to calculate the overall "Stat" rating.
|-|Special Abilities & Traits=
Individual ships can differ widely in usefulness due to special abilities or other attributes.<br/>
These can range from extra slots to special AACIs, speed differences, Touch mechanics, special equipment, or even extra range (which can be both good or bad).<br/>
As each of these traits is generally similarly felt across all classes, these are given flat values added to each ships' total "Bonus" rating.<br/>
{{User:Memetails/Specialty Bonus Values}}
|-|Other Bonuses=
Additionally, each ship can be awarded bonus points based on the number of ships that exist in that particular class.<br/>
* For example, Akashi, the only Repair Ship, receives a large number of extra points in comparison to a Destroyer, which have over 100 ships in their class.
* The bonus points are calculated as: <math> \text{Bonus Points} = \log_{NumOfShipsInClass + 1} ( TotalNumberOfShips / NumberOfShipsInClass )</math>
|-|Results=
The following table is the final results for my personal fleet.<br/>
{{User:Memetails/Ship Rating Results}}
Notes:
* '''Included:''' Modernization for HP/ASW/Luck
* Marriage
** '''Excluded:''' Stat gains for LoS/ASW/Eva
** '''Excluded:''' Reduced consumption of Fuel/Ammo
** '''Included:''' Extra HP & Luck
</tabber>
→Tabber
<tabber>
|-|Methodology=
Just how much more important one stat is than another is hard to determine, especially when considering different map scenarios that could strongly encourage one stat over another.<br/>
However, the overall rating of a ship could be estimated with some (somewhat) basic math.<br/>
Each ship class is separated into their basic classes instead of separately, as most sub-classes perform largely similar roles (such as CA & CAV under the CA umbrella).abilities/traits that differ from the norm of their class.
Every ship is scored based on its stats, special attributes/abilities, and class.<br/>
* These small differences are instead handled by awarding points to each ship based on special
** For example, CAVs and Zara/Pola get bonus points for the ability to carry Seaplane Fighters.
* Each class also might place more or less importance on an individual stat.<br/>
** With this in mind, each stat/ship class combination can receive a scaling modifier via a stretch/compress multiplier to each stat on a class-by-class basis.<br/>
|-|Averages=
Each basic class has its stats averaged across the entire class, using the stats from each ships' final remodel.<br/>
* Special versions of the same ship are factored in separately from each other.
** For example, Kaga Kai Ni and Kaga Kai Ni Go both apply separately to the overall average.
|-|Stretch/Compress=
Each basic class also has a stretch/compress multiplier based on an estimation of how important an individual stat is to that class of ship.
* In most cases, stretch/compress values remain at '1' due to being universally important and/or there is enough variety between individuals to already have meaningful differences
** A value <1 indicates a 'Stretch', meaning the stat is '''less''' valuable
** A value >1 indicates a 'Compression', meaning the stat is '''more''' valuable
* Exceptions:
** Fuel & Ammo
*** No scalar multiplier
*** Instead, they have additive modifiers to the StdDev for each class
** '''Ignored:''' HP, Luck, LuckMax
|-|Standard Deviation=
The Stretch/Compress value is applied (via multiplication) to the Standard Deviation of each respective stat for each respective class.<br/>
This Standard Deviation value denotes how wide a standard 'step' is from the average for the class which directly factors into the rating clusters.<br/>
The following table displays the Averages, Standard Deviations, and Stretch/Compress values for each class.<br/>
{{User:Memetails/Averages Deviation and StretchCompress}}
|-|Stat Evaluation Clusters=
Using the above values, each individual ship is evaluated into clusters based on their stats.
The math & clusters are as follows:
{| class="mw-collapsible mw-collapsed wikitable typography-xl-optout" style="width: -moz-max-content; width: -webkit-max-content; width: max-content; text-align: center; vertical-align: middle; "
|-
! Cluster
! '''DistanceFromAvg''' = ShipStat - ClassAvg
|-
! Insane:
|> 2 * (StdDev * Stretch)
|-
! Outstanding:
|> 1 * (StdDev * Stretch)
|-
! High:
|> 0.5 * (StdDev * Stretch)
|-
! Normal:
|< 0.5 * (StdDev * Stretch) '''&''' > -2 * (StdDev * Stretch)
|-
! Low:
|< -2 * (StdDev * Stretch)
|-
! Bad:
|< -3 * (StdDev * Stretch)
|-
! Terrible:
|< -4 * (StdDev * Stretch)
|-
| colspan="2"; style="text-align: left;"| Note: 'Stretch' refers to both Stretch & Compress
Clustering Exceptions:
* Fuel & Ammo: Additive Stretch/Compress instead of multiplicative, table is basically inverted due to lower = better
* Luck & LuckMax: Set values shared across all classes.
**Insane/Outstanding/High/Normal/Low/Bad/Terrible = > 50/35/25/10/5/0/NA
* HP: Instead based on 'good/bad' values via the overkill mechanic, such as %4, %6, %12, %4-1, etc.
|}
|-|Stat & Rating Multipliers=
Each stat is assigned a numerical multiplier based on an estimation of total value/importance.<br/>
Each cluster is also assigned a numerical multiplier based on an estimate of how good or bad being in that cluster is.<br/>
These 2 values are then multiplied together, resulting in the following table:
{{User:Memetails/Stat & Rating Multipliers}}
For example:
* Firepower is generally more important than LoS: '''Firepower > LoS'''
* Night power is generally more important than Firepower: '''Night > Firepower'''
* It can be concluded then that: '''Night > Firepower > LoS'''
* The 'Normal' cluster is left out of the table as it has a value of 0, being neither good nor bad
Using this table, each ship can then add together the value for each stat/cluster combination it's part of to calculate the overall "Stat" rating.
|-|Special Abilities & Traits=
Individual ships can differ widely in usefulness due to special abilities or other attributes.<br/>
These can range from extra slots to special AACIs, speed differences, Touch mechanics, special equipment, or even extra range (which can be both good or bad).<br/>
As each of these traits is generally similarly felt across all classes, these are given flat values added to each ships' total "Bonus" rating.<br/>
{{User:Memetails/Specialty Bonus Values}}
|-|Other Bonuses=
Additionally, each ship can be awarded bonus points based on the number of ships that exist in that particular class.<br/>
* For example, Akashi, the only Repair Ship, receives a large number of extra points in comparison to a Destroyer, which have over 100 ships in their class.
* The bonus points are calculated as: <math> \text{Bonus Points} = \log_{NumOfShipsInClass + 1} ( TotalNumberOfShips / NumberOfShipsInClass )</math>
|-|Results=
The following table is the final results for my personal fleet.<br/>
{{User:Memetails/Ship Rating Results}}
Notes:
* '''Included:''' Modernization for HP/ASW/Luck
* Marriage
** '''Excluded:''' Stat gains for LoS/ASW/Eva
** '''Excluded:''' Reduced consumption of Fuel/Ammo
** '''Included:''' Extra HP & Luck
</tabber>