Application Of K-modes Clustering Algorithm And Greedy Algorithm Based On Multi-attribute Weight In Dormitory Allocation

In recent years,the number of students in China’s universities has increased dramatically due to the annual expansion of university enrolment.As the post-00 s with distinctive personality traits gradually enter university campuses,college students’ demand for personalized dormitory allocation is also increasing,and the traditional manual dormitory allocation method seems inadequate.This thesis aims to develop an automated dormitory allocation model based on students’ personality traits,using the kmodes clustering algorithm and greedy algorithm to assign students to dormitories,so that students with similar personalities can be grouped and students with different personalities can be separated from each other.The focus of this thesis includes two main areas of research work.(1)Improvements to the traditional k-modes clustering algorithm.The traditional k-modes clustering algorithm is simple and fast,but there are limitations in both the measure of dissimilarity of classified data and the selection of initial cluster centroids.This thesis introduces the concept of attribute value weights to improve this algorithm and proposes a k-modes clustering algorithm based on multiple attribute weights.To measure the dissimilarity of classified data,a new dissimilarity measure formula is defined by combining internal dissimilarity(the difference between the attribute values themselves)and external dissimilarity(the distribution characteristics of the attribute values in the data set).For the selection of the initial clustering centroids,data objects with low outliers and far apart from each other are selected as the initial centroids by combining the outlier and dissimilarity.The improved algorithm is tested on the UCI dataset and the experimental results show that the algorithm performs well.(2)A dormitory allocation model is proposed based on the improved k-modes clustering algorithm and the greedy algorithm.In this thesis,the important personality factors affecting interpersonal relationships in dormitories were identified using a questionnaire and used as the main attribute characteristics of students in the dormitory allocation process.The improved k-modes clustering algorithm is then applied to the dormitory allocation model and the greedy algorithm is used to reallocate the remaining students after the clustering is completed.The modeling was completed and applied to the questionnaire dataset for analysis to verify the feasibility of the model.By means of simulation experiments,the model is compared with other dormitory allocation models in terms of model operation time and model allocation effect as the number of students increases,and the practicability of the model is verified.