The Research On Classification Method And Application Based On Neighborhood Rough Set And SVM

In the era of big data,data analysis is to discover the internal connections between things through the analysis and processing of massive data,thereby mining valuable information hidden in the data.However,it often involves high-dimensional data,and it is necessary to effectively eliminate redundant features and noise data in the dataset to obtain high-quality reduced dataset.Neighborhood rough sets can better handle the uncertainty of data in incomplete information systems,derive the minimum attribute reduction set of knowledge systems,and maintain the classification ability unaffected.At the same time,with the rapid development of computer technology,relevant technologies and methods in research fields such as machine learning and data mining have been widely applied to the field of education.This thesis combines neighborhood rough set,sparrow search algorithm,and support vector machine to study,and carries out classification prediction analysis on middle school student performance.The main research results are as follows:(1)Aiming at the fact that each attribute in a neighborhood rough set has the same weight,and each attribute has a different degree of impact on decision-making,an attribute reduction algorithm based on weighted neighborhood rough sets is proposed.The algorithm firstly uses the CRITIC weighting method to weight conditional attributes,and introduces a weighted distance function to calculate neighborhood relationships to obtain weighted neighborhood relationships.Secondly,construct weighted neighborhood rough sets,evaluate the importance of subsets using attribute dependency,and perform attribute reduction to find the optimal attribute subset.Finally,10 datasets from the UCI database are used for experimental analysis to compare the performance with four different attribute reduction algorithms.The experimental results show that the algorithm can not only obtain the minimum attribute reduction set but also ensure the classification accuracy of the reduced data.The algorithm has effectiveness and practical application value.(2)Aiming at the fact that most attribute reduction algorithms in neighborhood rough set are designed based on uncertainty measures,there may be attributes in the resulting reduction set that can only divide a small portion of the samples correctly,which increases the computational complexity and is prone to over fitting problems.An attribute reduction algorithm for neighborhood rough sets is proposed.This algorithm further prunes the reduction set obtained using the classic attribute reduction algorithm.By calculating the importance of attributes in the initial generation reduction set,the attributes with the smallest importance are sequentially deleted,and then the classification accuracy is compared,eliminating those attributes that cause the accuracy to decrease,making the resulting reduction set more efficient and with lower dimensions.Experimental results show that the algorithm has better reduction performance.(3)A classification method based on neighborhood rough sets and SVM is proposed and applied to the classification and prediction of middle school students’ grades.Firstly,the sparrow search algorithm is used to optimize the selection of support vector machine parameters,and the classification accuracy is used as the fitness value of the sparrow population.The positions of discoverers,followers,and observers are continuously updated to find the optimal parameter combination.Secondly,the student achievement data set is reduced by using an attribute reduction algorithm of neighborhood rough set proposed in(2)to obtain the feature subset,and the reduced data set is input into the optimized support vector machine for classification prediction.Finally,through experiments to compare and analyze the accuracy,F1 score,and AUC value of this classification method with five classification methods,it is found that the classification method proposed in this thesis has a good effect,and attributes with high importance are important factors that affect performance.