This project aims to assist with sorting your colonists into a better set of available jobs.
You will supply a list of which jobs your colony has available, and a list of your colonists. The program will then run several iterations of assigning them to various jobs and coming up with a pretty good set of assignments.
This project depends on at least python 3.8, and the modules in requirements.txt. To install the required modules, you should be able to run:
# Mac/Linux
python -m pip install -r requirements.txt
# Windows
py -m pip install -r requirements.txt
The workertypes.csv file contains the different types of jobs available in MineColonies, with the exception of Library Student and Pupil. Because the sole purpose of those jobs is to raise that particular stat, it would not be a good idea to optomize for putting a highly skilled person into that job. :) Raise your colonists skills at the Library or School separately from this sorting.
The current job list was scraped from the MineColonies worker system page on the wiki 1.19.2 version.
I've felt that some worker types are more important than others, so there is a weight component for each type. Set these all to 1.0 if you want to treak each type the same. For example Archers and Builders have the same skill requirements, but I would rather have highly skilled builders. So I weight Builders a little higher, and Archers lower. Feel free to adjust these weights in workertypes.csv.
The input for the Jobs list is a simple CSV file with just two columns, "Jobname" and "Count". List each type of job you have available in your colony along with the count. You don't have to specify every job if you don't have one, or you can set the count to zero. See the example jobs.csv file.
The input file your your colonists is a CSV file with the following column names. Order does not matter. - Name - Athletics - Dexterity - Strength - Agility - Stamina - Mana - Adaptability - Focus - Creativity - Knowledge - Intelligence
If you have your three input files, and you don't want to change any defaults, you can run the program with the following:
~/mcsort $ python sort.py
Score 370.4 |████████████████████████████████| 5000/5000 100%
Score 370.4 |████████████████████████████████| 5000/5000 100%
Score 370.4 |████████████████████████████████| 5000/5000 100%
Score 367.4 |████████████████████████████████| 5000/5000 100%
Score 370.4 |████████████████████████████████| 5000/5000 100%
JobName Worker Primary Secondary
---------- ----------------- --------------- ---------------
Blacksmith Samuel Q Ashton Strength 27 Focus 20
Builder Ellianna W Jones Adaptability 27 Athletics 20
Builder Hana H Gedding Adaptability 26 Athletics 19
Carpenter Kiaan C Andrews Knowledge 28 Dexterity 22
Cook Remi P Greene Adaptability 25 Knowledge 19
Courier Bo A Gisborne Agility 27 Adaptability 25
Courier Kasem N Collard Agility 25 Adaptability 31
Forester Gloria G Addicock Strength 30 Focus 17
~/mcsort $ That will kick off the trials and show you the best score from all the trials.
This program uses a simplistic version of an evolutionary algorithm to sort colonists in to jobs in an attempt to find a 'pretty good' set of assignments where the overall skills of the colonists are well utilized. I don't say 'best' in any case here, as the randomized nature of the algorithm makes it difficult to say any given set of job assignments is the abolute 'best'. And 'best' is highly subjective for each player and set of colony restrictions.
One trial consists of first assigning your available colonists to random jobs. It does't matter if you have more colonists than jobs, or more jobs than colonists. It should do the best it can with what's available. A given set of jobs with colonists assigned I call a Job Set.
Once the initial Job Set has been generated, it is scored. The scoring function is pretty basic. It goes through all the jobs in the Job Set, and adds up the weighted skill values for the primary and secondary skills of that Worker Type. Because the Primary Skill for a given Worker Type is more important, it has a weight of 1.0, and the secondary skill has a weight of 0.7. The skill of the colonist assigned to the job is multiplied by the weight of that skill, then are added up across all job/skill/colonist combinations. The result is the score for this Job Set.
Once the initial score is found, a set of swapping iterations happens. One occupied job is randomly selected, and then a different colonist is randomly selected. If both colonists are employed, then their jobs are swapped. If the second colonist was unemployed, then that colonist is assigned to the frist job.
Once a swap has happened, the new Job Set is scored. If this Job Set has a higher score than the previous one, it is promoted to the 'current best' Job Set, and the iterations continue.
By default each trial runs 5000 generations.
Through emperical testing, I have found that about 5000 generations seems to arive at the best score that trial is going to get. Running it out to hundreds of thousands of generations doesn't seem to budge the score. However due to the randomized nature of the initial Job Set, successive trials can result in measurable improvements in the final score. Therefore I have chosen to run multiple full trials, picking the best out of all the trial results. By default there are 20 trials run.
Many of the defaults can be changed with command line options. See python sort.py --help for details.
usage: sort.py [-h] [-t TYPESFILE] [-j JOBSFILE]
[-c COLONISTSFILE] [-n NUMTRIALS]
[-g GENERATIONS] [-p PRIMARYWEIGHT]
[-s SECONDARYWEIGHT]
optional arguments:
-h, --help show this help message and exit
-t TYPESFILE, --typesfile TYPESFILE
Worker Types input CSV file
-j JOBSFILE, --jobsfile JOBSFILE
Jobs input CSV file
-c COLONISTSFILE, --colonistsfile COLONISTSFILE
Colonists input CSV file
-n NUMTRIALS, --numtrials NUMTRIALS
Number of Trials to run
-g GENERATIONS, --generations GENERATIONS
Number of Generations per Trial
-p PRIMARYWEIGHT, --primaryweight PRIMARYWEIGHT
Weight of the Primary Skill
-s SECONDARYWEIGHT, --secondaryweight SECONDARYWEIGHT
Weight of the Secondary Skill