校務研究之學生學習成效分析

巨量資料分析話題持續延燒，該如何運用資料分析讓資料進行加值，是目前的一個重要議題，不僅僅只是產業界欲利用資料加值創造競爭優勢，如今因社會環境逐漸變遷，少子化現象成為未來隱憂，同時大專院校自主權增加，需自行承擔招生壓力，該如何將有限的校務資源達到最妥善運用，是大專院校目前需要克服的問題。國外利用資料分析進行「校務研究」（Institutional Research）目前已行之有年，目的為分析校園內、外部資料，將資源使用率最大化或提供決策資訊，這股風氣漸漸蔓延至國內，國內開始有大專院校發展校務研究資料分析平台。在現有校務研究資料分析平台中，常見於分析課程與學生學習成效，但學生學習成效至目前為止，定義相當模糊，沒有統一基準進行評估。有鑑於此，本研究針對學生學習成效部份，試圖定義出學生學習成效之落點族群，目標在於發現可能需要接受輔導之學生，並找出影響學習成效之關鍵指標變數，以供後續區別預測與追蹤，本研究以C大學資訊管理系為例。本研究使用分群的方式，區分出學生之學習效果，再對學習效果進行預測，可準確預測出可能需要接受輔導學生群組。根據研究結果顯示，前一學期的成績類變數，具有高度關聯性，其中以大二下學期時，上學期資管課程平均影響最甚，使用該變數預測結果最為精準。

關鍵字

校務研究；學習成效分析；資料探勘；區別分析

並列摘要

Big Data continues popular. The issue pertaining to ways of utilizing data analysis for value-added data becomes important, which is not just because industries intend to make use of its creating competitive advantages, and current social environment changes gradually that declining birthrate is worthy of worry in the future, but also that the autonomy of universities and colleges enlarges which make them bear enrolling pressure on their own. In light of this, how to make the most proper use of limited school resources to achieve the most appropriate use of institutional resources becomes an issue needing universities and colleges to overcome. These of data analysis for "institutional research" has been implemented for years, which aimed at analyzing campus data, external data, and maximize resource utilization or provide decision-making information. This trend spreads gradually throughout Taiwan, and universities and colleges of Taiwan began to develop institutional research data analysis platform. Many of the current platform of institutional research target on the analysis of curriculum and student learning effectiveness. However, as of now, the definition of student learning effectiveness is quite vague without uniform benchmarks for further evaluation. For this reason, this study intended to define target group in evaluating student learning effectiveness, aimed at finding out students who need counseling, and identifying key indicator variables among for subsequent differentiated prediction and tracking. Department of Information Management, C University was served as the subject in the case study. Finally student’s learning effectiveness was clustered and distinguished and further for prediction. From these, we may predict student groups who might need counseling accurately. According to the findings, in the last semester, effectiveness variables had high degree of correlation, among of which, in the 2nd semester as a sophomore, the average score of the course titled Information Management was affected to the most, where prediction via such variables generated the most accurate effectiveness.