Graduate Employment Forecast On Gaussian Mixture Model And Integrated Learning

The employment orientation and quality of graduates have always been a matter of great concern to colleges and universities.Employment consultation and counseling are also tasks that colleges and universities are very concerned about.Colleges and universities store a large number of graduates and employment trends every year.The task volume is very large and very difficult by artificially analyzing the rules in these data.It will also rely heavily on subjective factors.It is difficult for students to apply these data themselves.At the same time,tutors will also pay some attention to some of the characteristics of graduates during employment counseling.It is difficult to take full account of all the conditions of graduates.The years of personal experience of the tutor is not easy to replicate so that cannot be widely popularized.In the period of large demand,a few tutors are difficult to complete the tutoring needs of a large number of graduates.In response to these problems,this article first studies data clustering and prediction algorithms.According to the types of clustering algorithms,some common clustering algorithms are briefly described,and the principle of Gaussian mixture model is emphasized.Then the mathematical principles of BP neural network and support vector machine in classification prediction algorithm are studied.Then construct the indicator of the graduates’ situation,and construct a more complete description indicator based on subjective and empirical cognition,integrated forecasting situation,external economic environment factors,and correlation regression analysis.The complete data is normalized and out-of-order partitioned to construct a training set and a test set.Through the design of related prediction algorithms,a large amount of data is used to cluster analysis of graduates,and the Gaussian mixture model algorithm is used to realize the mining of potential categories of graduates,and it is reasonable.The future employment factors of graduates are analyzed using the integrated algorithm that integrates BP neural network and support vector machine model,and a better predictive model parameter scheme is obtained based on comparative experimental research.Through comparative research,the performance of individual classifiers and ensemble models shows that ensemble models can better make up for the deficiencies of individual classifiers.The performance of adding clustering results indicates that the clustering results can promote the increase of classification accuracy.The performance of data balancing indicates that data balancing can improve performance.The finally formed classifier performs well on the test set,and can reach prediction accuracy of 84.76% and 85.35% in predicting the nature of the work unit and predicting salary levels,respectively.This research provides a method that combines data mining and analysis methods with graduate employment issues and provides more research ideas for counselors who study graduate employment issues.The results of the clustering algorithm in this research can provide graduate category labels.For different predicted categories,it can provide graduates with possible future employment conditions based on their own situation,and provide employment reference information and personal growth plans for graduates or other students at school.