A Memetic Genetic Algorithm Framework for Public Transit Route Optimization: The Genetic Scissors Method

This study proposes a memetic genetic algorithm framework, termed Genetic Scissors, for improving the spatial directness of public transit routes based on the rationality measure. The framework is formulated on an extended problem structure and incorporates a same-location constraint that prevents stops sharing identical geographic coordinates from being selected consecutively within the same route, together with an enhanced repair procedure that preserves the positional integrity of route start and end stops. It combines three components: a two-phase fitness structure that resolves the scale incompatibility between absolute distance and normalized rationality, six original intensification operators acting at the node, segment, block, and in-route evolutionary search levels, and a regression-based adaptive parameterization scheme responsive to problem size. Experimental analyses on 213 normal clusters derived from the Istanbul public transit network, together with evaluations on a 109-cluster fair-comparison subsample, show that the proposed method improves the existing routes in 81.7% of cases, with a mean rationality improvement of 2.41%, and yields a highly significant Wilcoxon signed-rank effect ( p < 0.001 , r = 0.750 ). In direct comparison, Genetic Scissors outperforms the reference method in 84 of 109 clusters ( p < 0.001 , r = 0.714 ).