Optimizing task scheduling with a genetic algorithm in Python

This article develops a software module for optimizing a project schedule using a genetic algorithm written in Python. The limitations of classical planning methods, such as critical path method and program evaluation and review technique, are analyzed in the context of multi-criteria and high uncertainty. The application of evolutionary computation to solving the non-deterministic polynomial-time hard problem of minimizing the total project completion time is explored. A formal problem statement is developed, including an objective function, a set of constraints, and a chromosome structure. Key operators of the genetic algorithm are presented: population initialization, tournament selection, single-point crossover, and mutation, adapted for working with time parameters. Computational experiments on test data are conducted to evaluate the convergence and quality of solutions. It is found that the implemented algorithm allows reducing the duration of the test project from 50 to 6 days. A comparison of the results with the critical path method is formed, demonstrating a significant advantage of the genetic approach. The practical significance of the development for automating planning in project management, construction, and the information technology sector is substantiated.