基于遗传聚类算法的出行行为分析

针对K中心点聚类算法对分类数据聚类的有效性和遗传算法良好的自组织、自适应和自学习能力，提出了基于遗传聚类算法的出行行为分析方法。该方法采用整数编码，用活动模式间的匹配度度量模式对象之间的相异度，以各活动模式与最近聚类中心点之间相异度的总和为适应度函数，探讨了K中心聚类与遗传算法相结合完成分类对象聚类分析的方法；通过算法在不同数据量和不同参数设定下仿真结果的比较，提出了关键参数的推荐值。研究表明，新方法不仅能很好地解决孤立点和局部最优的问题，同时还提高了算法的收敛速度，降低了计算成本，能很好地解决分类数据的

Travel behavior analysis using genetic clustering algorithm

Based on the good performance of K-medoids clustering algorithm for categorical data and the nice self-organization, self-adaptation and self-learning of genetic algorithm, this paper aimed to develop a methodology for the clustering of activity patterns with a genetic algorithm based clustering method. The proposed method used integer coded chromosome. The dissimilarity measure between two activity patterns was defined as the total number of mismatches of activity types at a corresponding time index and the fitness function was defined as the sum of dissimilarities of all objects to their nearest medoids. The results for different sizes of data sets and for different parameter settings were compared and based on this recommended parameter settings were provided. It is demonstrated that the algorithm is good at preventing premature convergence, decreasing the sensitivity to outliers and that it is fast converging and is a good solution for categorical data clustering analysis.