使用者行為數據分析：以線上學習歷程紀錄分群為例

Abstract

數據分析已是現代商業不可或缺的一環，企業也開始蒐集使用者行為數據以作為其商業決策的基礎。但由於企業難以建立兼備產業知識與數據分析能力的團隊，亦難以評估導入效益，導致企業缺乏導入數據分析的意願，即便勉強執行也常導入不順。 本文以線上教學平台導入分群模型作為產品開發基礎為例，詳列企業導入數據分析的過程與挑戰，以作為企業規劃的參考。本文詳細記錄如何使用平台學生端的線上行為數據，透過實際進行數據清理篩除不必要與錯漏的資料，並以敘述性統計方法進行分析後選擇並建立學生特徵值，最後使用簡單統計量分群方法成功建立分群模型，並於教師端與企業端皆取得正面的回饋，成為下一階段產品開發的基礎。 此個案研究驗證了使用者行為資料分析具備提高企業價值的能力，並透過此個案研究記錄了完整資料分析流程與相關實作細節，最後提出企業初期導入及擴大實行的建議。

Data analysis has become the most important function of a modern business, and companies also begin collecting user behavior data for their business decisions. However, it is too difficult to recruit a team with both domain knowledge and data analysis capabilities, so the companies are hard to evaluate the benefits of introducing data analysis. Because the companies have few interests to introduce data analysis, and even if they are trying to do so, they often fail. This article takes an online education platform as an example of implementing data analysis in a company. The results of this case were eventually turned into the foundation of product requirements for their future iteration. The article records in detail how to use the online behavior data of students on the platform, removes unnecessary and erroneous data through the data cleaning process, and analyzes it with descriptive statistical methods to select the student features. At last, a clustering model was successfully built by using simple statistical values, and the model got positive feedback from both the teacher and the enterprise. This case study assures that user behavior data analysis is able to improve business value, and records the complete process and implementation details for reference. Finally, the case study concludes with suggestions for the early implementation and the expansion when enterprises introduce data analysis to their business processes.