The Application Research Of Decision Tree Algorithm In The Teaching Quality Evaluation System Of University

Nowadays, many domestic universities take advantage of IT (Information Technology) to improve its management. The school runs all kinds of systems and databases, such as academic information management system and employment information management system. The long-term running of these systems accumulates a large amount of data. But because of the lacking of information knowledge and technology, managers just acquire superficial information through simple statistics or sorting, leaving more information for further exploration.This thesis attempts to put the Decision Tree and Association Rules into the use of the construction of the Teaching Evaluation System in colleges and universities, with the hope of explore the relationship among teachers'personal factors, teaching operation factors as well as the teaching effect, thus provide basis for the teaching evaluation. The thesis studies various aspects that affect teaching quality--subjective factors of teachers (position, educational level, age, gender, teaching methods, etc.) and the objective factors of the teaching environment (nature of curriculum, course type, enrollment size, workload of each week). It combines these factors together and makes comprehensive analysis of related factors.First of all, the thesis makes summary of the basic knowledge in data mining, and introduces the Decision Tree and classic Association Rules, and makes a brief overview of the Microsoft SQL Server 2005 Analysis Services. Taking the teaching evaluation data of a college in 2009-2010 as an example, the thesis uses the Microsoft SQL Server Analysis Services as a tool to show the whole process of data mining, which including pre-processing technology such as data cleaning, data conversion and data integration. It uses the Decision Tree to build the forecasting models for teaching evaluation. Through the calculation of information gain, the thesis finds out the effective Association Rules. Based on the data of teaching evaluation, the thesis makes use of the methods of Association Rules to analyze the association between the input attributes and the evaluation results, and then makes analysis the association of each input attribute and the evaluation result, and finds out the Frequent Itemsets. By setting the Support Threshold and Confidence Threshold, it finds out Frequent Itemsets to meet the support and confidence. This provides verification for the rules of Decision Tree and finds out the key factors that affect teaching. The conclusion part makes a detailed analysis of the experiment result, finds out the main problems and proposes countermeasures for teaching improvement, aiming at carrying out a reference for the advancement of teaching.