Analysis Of Software Engineering Courses Based On Bayesian Network

Nowadays,following the fast development of domestic higher vocational education,examination as a kind of test tool to students study level plays an important role in the teaching process,but the traditional teaching methodology just focus on the scores,it cannot analyze the deep reason behind students,this way cannot satisfy the new need of modern classes any more,thus,through the way of analyzing the collected sample data,this paper tries to propose an new way to evaluate the effect of teaching and learning,hopes that the analysis would provide teaching management with targeted feedback information,so as to make effective observation and evaluation in study behavior of students.This paper used Bayesian network as the data basis to analyze the current influence factors of test results,Bayesian network as a kind of effective uncertainty knowledge representation and deduction tool,has been widely used by many researchers in the field of uncertain knowledge discovery.A Bayesian network is composed of nodes and link nodes to edge,node can represent a factor,a variable or a hypothesis,directed edge represents the connection between various nodes,each node is equipped with conditional probability parameter table,which used to quantify the degree of interdependence among nodes.The examination results of software engineering testing course were used as the breakthrough point in the analysis.First choosing various influence factors of performance as the nodes,and correlation among factors as the directed edge to build a Bayesian network.The influencing factors should be filtrated reasonably,and then the relevant data were collected,sorted and discretized.After discretization,on the basis of the data,using structure learning algorithm to construct a Bayesian network,then studied the parameter to get the prior probability of each node,and finally used Bayesian network to deduct,in order to confirm the correlation between test scores and influencing factors.Meanwhile,dig out the potential relationship among the various factors,which obtained a certain referential valuable conclusion.All experimental works of this paper were used FullBNT software package under MATLAB environment,through the experimental analysis,obtained data which including gender,poverty level,and leading factors such as students software course scores and school performance,and put forwarded personal constructive and feasible opinion and the suggestion to current teaching designing,teaching organization,and teaching management.The purpose is to use kinds of data experiments to support the teaching process,following the continuous improvement in the aspect of technical data analysis,hope to find out some disadvantages in the traditional teaching methodology,gradually build up a set of new methodologies which are totally based on data collection and analysis,any new adjustment in school teaching should have data as the strong support basis,if the opinion in this paper could make some helpful attempt to university teaching methodology and the evaluation of students,it would be a great honor to me.