| Summary: | 碩士 === 雲林科技大學 === 資訊管理系碩士班 === 97 === The important period of students for training their study habits and personality is elementary school stage. In order to find the related impact factors of student achievement, and achieve a balanced development for moral, intellectual, physical, social and aesthetic. Therefore, parents and teachers must improve the learning conditions in which children can have the more balanced development. This dissertation uses scientific methods to validate the generalized rules which can provide decision makers as a reference and make the principle of education.
In this study, the dataset is practically collected the score of graduated students with 670 records from “Nantou County C elementary school” in 2006-2008. The dataset include two dimensions with 15 attributes: (1) Study efficiency dimensions: the average score of language, mathematics, nature, art and humanities, social, health and physical education, integrated activities, and day-to-day behavior performance, (2) Environment and background dimensions: parental education, parental occupation, parental age, number of children, the ordering of family member, the identity and background of students, and mentor. This research combines RST and selecting attributes as mainly research methods, and uses three kinds of other data mining methods (Naive Bayes, Multi-Layer Perceptron, Decision Tree) for comparing with proposed method. The results in this research are listed as follows:
1. The accuracy is 90.24 % by using RST method, after selecting attributes by OneWay ANOVA, the accuracy is 90.46 % ,and then selecting attributes by Machine Learning, the accuracy is 90.66 %.
2. The rules of achievements are extracted by RST, calculate the ranking of these attributes is: mathematics, nature, language, art and humanities, integrated activities, health and physical education, the identity and the background of students, day-to-day behavior performance, parental education, social, number of children, the ordering of family member, parental age, parental occupation, mentor.
3. Compared with other data mining, the RST method has the advantage of easy understanding.
|
|---|
