結合機器學習技術之數位遊戲式學習模式對國小生數學學習之影響

Effects of a Machine Learning-Based Digital Game-Based Learning Approach on Elementary School Students’ Mathematics Learning Performances

近年來，數位遊戲式學習在教育領域受到相當的矚目，影響層面廣泛。但許多研究發現，在遊戲中若能結合學習策略或教學支持，將更能提升學生的學習成效與動機；另一方面，機器學習技術日新月異，其中的決策樹演算法可快速進行資料的分類整理，並具有分析預測的能力。因此，本研究設計出一個結合機器學習技術的教育遊戲，利用機器學習中的決策樹演算法進行動態偵測，於遊戲過程中即時預測學習困難的學生，並給予適當的學習輔助。本研究採用準實驗研究法，研究對象為國小四年級共60位學生，實驗組以結合機器學習技術的數位遊戲式學習模式進行學習，控制組則以一般數位遊戲式學習模式進行學習。實驗結果發現，實驗組在學習成就上顯著優於控制組。此外，在數位遊戲中結合機器學習技術，可有效提升學習者的學習態度與自我效能，幫助學習者進行有效的學習。

In recent years, digital game-based learning has made a splash in the field of education and has had a wide range of impacts. However, most studies have found that the learning achievement and motivation can be improved when games are combined with some learning strategies or instructional support. The decision tree in machine learning technology can quickly categorize and organize data and make analysis and prediction. Therefore, this study designs an educational game that integrates machine learning technology to provide appropriate learning assistance to students with learning difficulties during the game by using the machine learning motion detection function. The experimental group was conducted in a digital game learning mode with machine learning technology, while the control group was conducted in a general digital game learning mode. The results showed that the learners in the digital game learning mode with machine learning technology were significantly better than the learners in the general digital game learning mode in terms of learning achievement. In addition, the combination of machine learning technology in digital games can effectively enhance learners’ learning attitude and self-efficacy, and help learners do well in learning.