A Study of Teacher’s Teaching Assessment by using Data MiningTechnology

• Main Authors: Ting Huang, 黃婷
• Other Authors: Chih-Li Wang
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/81146233817997826681

碩士 === 銘傳大學 === 應用統計資訊學系碩士班 === 101 === With the popularization of higher education, Taiwan’s the previous elitist education to the popularised education, and university education is no longer the highest diploma. According to the Ministry of Education (2012) indicated there are about 162 colleges and universities in Taiwan. Because the double impact of declining birthrate and schools’ growth, so that making every universities faced insufficient enrollment, and they don’t have the advantage of other issues. So they begin to face their types and to improve the quality of teaching. That produces a set of effective assessment mechanism. In this study, we use the results of M University teaching evaluation’s database in recent years, and we use the data mining to identify the impact of teachers'' teaching evaluation factors. Also the results could used to improve teaching methods, improve teaching quality reference. In this study, we use working teachers from 95 to 100 academic years in M University as the researchful target. And we use the decision tree for C5.0 and CHAID, neural networks and multicategory logistic regressions of data mining’s technologies, to identify the impact of teacher performance evaluation factors and using evaluation chart and classification matrix to determine the four mining technologies in teaching evaluation database’s useable solution. Finally we find C5.0’s accuracy is the highest accuracy in the opening teacher’s characteristic. The multicategory logistic regressions and neural network are the highest accuracy in opening teaching’s quality. And the CHAID’s accuracy are the most lowest in these two database of opening teacher’s characteristic and opening teaching’s quality.