Flight Scheduling EA

Description

An evolutionary algorithm for solving the combinatory optimization scheduling problem of writing daily flight schedules.

Note: Simulated flight data is used for releasability.

Installation:

1. Ensure Python 3.6 is installed.
2. Clone the repository.
3. In the console, inside the top level project directory:

python3.6 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Run the Program

Consider expanding your terminal horizonally, the results table is often wider than the default terminal width.

From the top project directory, in the console:

python run.py

Change the way the program runs:

• To change the schedule used in the program, in run.py, change the number in the event file name to another number between 1 and 5.

-from app.state.events1 import events
+from app.state.events3 import events

• To change any of the parameters, just edit the values in ./app/ga/parameters.py. Values chosen for the GA are in comments beside the values for reference if changed..
• To see the hamming distance at the conclusion of the run, add the following parameter to the call to main() in run.py (This is time consuming).

print_hamming=True

• To view the plotted results, navigate to ./tmp. Images of the figures will be saved there.

Directory Structure

• app - everything that matters
  • analysis - diversity caluclation and plotting
  • constants - app-wide constants defined
  • ga - fitness calculations, genetic operators, parameters
  • models - definition of model classes, incl Constaint, Individual and Population
  • state - schedules, pilots and events
  • test - unit tests
  • tmp - fitness plots saved here