校務研究資料庫的建構與分析應用

DEVELOPMENT AND APPLICATION OF DATABASES FOR INSTITUTIONAL RESEARCH AND ANALYSIS

曾元顯(Yuen-Hsien Tseng)

《當代教育研究季刊》 24卷1期 (2016/03) Pp. 107-134

https://doi.org/10.6151/CERQ.2016.2401.04

摘要

本文根據教育科學、圖書館學、資訊工程等領域的知識，結合多年來處理校內外資料庫的實務經驗，以及對近年技術產品的瞭解，闡述校務研究資料庫建置的較佳實施概念，並分析比較各種建置方案的適用時機與優缺點，最後以三項代表性的具體實例說明本文提及之實施概念的綜效。具體而言，本文介紹採、編、典、藏、用五項可持久運作的資料庫建構作業流程與注意事項，說明資料正規化與反正規化的用處，透過概念驗證作業提供與國內廠商互動的經驗，並以實際校務數據的分析案例，分享應用經驗。整體而言，資料蒐整作業（即：採、編、典、藏）仍是最費時、費力的流程，一旦完成，後續的分析運用便容易進行。目前的視覺化分析工具將越來越便利，讓各類型使用者得以更有效率的從大量資料中發現特殊樣態、形成假說，進而對資料做各種查詢與探索，以獲得具體事證支持決策。展望未來，除了視覺化工具越受依賴外，事件演進模擬技術，也將扮演重要角色，其可讓使用者事先知道各種因素變化後的最終結果，讓分析平台更具價值。

關鍵字

資料一致性 ； 資料正規化 ； 資料素養 ； 資料倉儲 ； 視覺化分析

並列摘要

This article elaborates on the possible best practice of developing databases for institutional research and analysis, based on the knowledge of Educational Science, Library Science, and Information Engineering, years of experience in developing educational databases, and a recent survey of related technology and products. Several developing options are compared to show their benefits and disadvantages under different conditions. Three representative analysis tasks are reported to verify and show the synergy of the mentioned ideas and experience. In particular, this article proposes a sustainable workflow: (1) data collection and aggregation, (2) cataloguing, (3) regulation, (4) archiving, and (5) usage, and describes their must-known caveats. The application situations of data normalization and de-normalization are described. Capability of domestic vendors of related products is briefly mentioned based on a proof-of-concept testing. And finally, real-world institutional analyses are conducted to share our experience. Overall, the first four processes in the above workflow are most time-consuming and costly. Once data have been well prepared, recent visualization analysis tools allow users to easily discover meaningful patterns and inspire hypotheses, and allow them to explore the database to find evidence to support their hypotheses and decisions. In the future, we expect that event evolution simulation techniques, which allow users to foresee the results given various input scenarios, could play an important role in educational data analysis, in addition to the maturing data visualization tools.

並列關鍵字

data consistency ； data normalization ； data literacy ； data warehouse ； visualization analysis

參考文獻

李欣宜（2015 年2 月17 日）。美國Top 4 技術長寶立明：大數據即將在五年內消失。數位時代。取自http://www.bnext.com.tw/。[Li, X. Y. (2015, February 17). United States top 4 chief technology officer Stephen Brobst: Big data is about to disappear within five years. Business Next. Retrieved from http://www.bnext.com.tw/]

紀馥安、許清芳(2015)。運用開放軟體R 處理大型教育資料庫。當代教育研究季刊。23(4)，121-153。

彭森明(2010)。大學生學習成果評量：理論、實務與應用。臺北市=Taipei, Taiwan:高等教育=Higher Education。

彭森明(2013)。高等教育校務研究的理念與應用。臺北市=Taipei, Taiwan:高等教育=Higher Education。

曾元顯(2014)。自動化資訊組織與主題分析近二十年來的研究與發展。教育資料與圖書館學。51(5)，3-26。