Research On Residents Selection And Optimization Model Based On KNN And Case-based Reasoning

With the development of the cities and the expansion of the new towns scales,the speed of replacement rate of the cities and towns continues to accelerate,making residents become one of the most dynamic elements of change on the map.At the same time,the requirements for map guarantees,such as military exercises,anti-terrorism drills,and emergency rescue operations,have become increasingly demanding,making the automatic synthesis of residential areas a key and difficult area for research."Selection" as the most important integrated operation in the synthesis of small and medium-scale maps has also become the focus and difficulty of research.At present,the selection method of residential qualification for establishing mathematical models for various metrics needs to be modeled separately when encountering mapping tasks.It has weak portability,high research and development costs,and lacks expert experience to support the diversification of map production.Studying more efficient methods for automatic selection of residential areas has become an urgent task for cartographic synthesis.This paper analyzes the research background and status quo of the selection of small and medium-scale residential areas.In light of the current lack of expert knowledge and guidance on the selection of small and medium-scale resident residential areas,this paper studies the KNN-based case selection model of residential based on the effective use of expert experience.At the same time,the case matching mechanism selected by residents is optimized,and the model optimization method based on iterative water injection principle and distance weighting,and the efficiency improvement method based on hierarchical processing and KD tree are proposed to further improve the accuracy and efficiency of KNN model.The main work and innovation of this article include the following points:(1)A case selection model based on KNN for residents' case is proposed.Aiming at the problems such as lack of expert knowledge guidance and imperfect case matching mechanism in the current resident selection methods,a KNN-based case selection method for resident inhabitants was proposed.Firstly,the results of the residents' interactions selected by the experts are taken as examples to build the case base after preprocessing;then,the case studies are conducted to calculate the similarity between the decision-making residential area and the case base;finally,the K highest similarities are taken.Nearest neighbors,statistics on the category of the K cases,the most number of categories as a result of case classification decisions.The selection model can effectively transform expert cases into decisions for unknown results and the accuracy of the decision is high.Compared with the decision tree method,noise is less affected,and an effective decision can still be made when the case size is small.It can learn the expert's comprehensive experience and imitate the purpose of the expert's comprehensive behavior.(2)A model optimization method based on iterative water injection principle and distance weighting is proposed.First of all,aiming at the insufficiency of attribute reduction and weight assignment methods in KNN algorithm,this paper proposes to adopt “Iterative Water Injection Principle” to improve and optimize the model.This method combines the traditional water-filling principle with the recursive feature elimination method to achieve the purpose of attribute reduction while minimizing the interference caused by redundant attributes to the case classification,and ensuring the accuracy and integrity of inference description information,and improving the accuracy of the KNN-based model for inferring case selection.Then,considering the primary and secondary impacts of K reference cases based on the KNN-based civic case inference selection model,a model of “distance-weighted” optimization method is proposed.By re-defining its discriminant function,the weight of the reliable case decision-making contribution is strengthened.To a certain extent,the sensitivity and dependence of the model on the K value are weakened,and the robustness and intelligence of the KNN-based model for case-based reasoning selection is enhanced.(3)A model optimization method based on hierarchical processing and KD tree is proposed.For KNN-based resident selection model of case reasoning,the efficiency of the model decreases after the scale of the case has been expanded.A model optimization approach is proposed that uses the case library to process and construct the KD-tree index.The program first classifies the case bases according to administrative levels,then builds a KD tree for each level of case bases,and finally implements case-based reasoning based on KNN at each level of the case base to obtain decision-making results for the decision-making residents.The program can effectively control the scale of case participation in the case-based reasoning cycle and improve the efficiency of the model.(4)The system experiment designed and implemented the automatic selection process of residential areas,integrated the algorithms and optimization functions proposed in this paper,and conducted systematic experiments to verify the validity and scientificity of the proposed theories and methods.