Analysis In Benchmarking Based On Data Mining And Teaching Management Process Research For Local College

Benchmarking is an effective method of performance management, which has been usedmore and more widely in many industries. Existing methods for this problems focus on thedirect application of traditional optimization model. Stochastic frontier analysis and dataenvelopment analysis are the most representative methods for benchmarking. Data mining iswidely applied in feature selection and extraction of data and research achievements areobtained. Hence, relevant data mining technology is advanced and applied to thebenchmarking, which can reduce the effect of bad data and subjective understanding foranalysis results, and can cater for the requirement analysis of complex situations. Due to thelate beginning and lack of comprehensive talent, existed research achievements could notmeet the need of practical work. Based on this, the method is rarely applied in performanceappraisal of local colleges and universities and teaching management.First, concepts and theories of the benchmarking are introduced. Classical data miningalgorithms are discussed.Applying the data mining method and benchmarking to managementof local colleges and universities is able to give an effective performance appraisal. This willplay an importance role in theory and method of teaching management. The major work andconclusions of present paper are following:The idea of fuzzy C-means clustering is discussed. The paper has established thebenchmark set optimal choice model based on FCM combing with demand of how to choosesuitable benchmark set in benchmark analysis. The method of GA is used to solve the model,and the algorithm complexity is analyzed.Concepts and theories of Principal Component Analysis (PCA) and Data EnvelopmentAnalysis (DEA) are discussed. We propose a hybrid models based on PCA and DEA. Itovercomes the shortcoming of difficulty precise analytic for many benchmark sets, and meetsthe needs of the performance appraisal.Indexing data of alternative benchmark sets are chosen by a university ranking and data of university development. We discuss how to choose an optimal benchmark set in two cases(actively benchmarking analysis and conservative benchmarking analysis) combing withproposed model.Performance is assessed of local colleges and universities based on DEA-WEI ModelPCA-DEA Model-proposed in this paper in the optimal benchmark sets. Experimentalresults show that achieves better results than DEA-WEI. So it provides the new data supportsfor improving the level of educational administration.