The idea of this project is to test what I can do with a Genetic Algorithm. To keep it general I used 'supply' and 'demand' as values, and both values have 'options'.
Supply: An object wanted by demand, this object has options.
Demand: A person seeking for an object with certain options.
We can now use this algorithm to distribute people (demand) among objects (supply) and match them by chosen options.
You can find a configuration file in the config folder with the following configurations:
| title | description |
|---|---|
| chance-of-mutation | chance of mutation per gene |
| pool-size | Size of population per generation |
| elitism | Transfer the parents to the next generation |
| max-stagnant | Max amount of stagnant |
Run the following command for creating test data:
$ php app ga:create-data [number-of-supply] [number-of-demand] [number-of-options]
This is for actually running the algorithm:
$ php app ga:run-algorithm
...
Generation: 813 (Stagnant:7) Fittest: 37/500
Generation: 857 (Stagnant:43) Fittest: 36/500
Generation: 871 (Stagnant:13) Fittest: 35/500
Generation: 896 (Stagnant:24) Fittest: 34/500
Generation: 918 (Stagnant:21) Fittest: 33/500
HALT! Exceeded 600 stagnant generations
[2020-10-13 19:43:09] Solution at generation: 1560 time: 62s
Genes: 473,348,228,144,462,312,165,234,323,84,125,264,215,351,390,461,29,373,435,424,395,384,160
Score: 30
-----------------------------
Generation: Current generation.
Stagnant: The number of generations went by without a better individual.
Fittest: The score of the fittest individual (0 is most fit).
- The algorithm does not work when you have less demand than supply