教育大数据背景下log数据挖掘与应用——以PISA(2012)中国区问题解决测验为例

Mining and Application of Log Data under the Background of Educational Big Data:Taking Problem-solving Test of China in PISA(2012) as An Example

log数据是教育大数据的一个子集,为分析问题解决过程提供了新的思路。文章截取PISA(2012)中国区的log数据,采用关系挖掘、聚类等教育数据挖掘方法 ,分析"交通"一题的答题时间、鼠标点击数,以及使用"有目的试误"策略情况。结果显示:(1)台湾和上海形成强烈反差,上海学生在高百分位上的解题时间和鼠标点击次数都较低,说明他们可能更在意时间成本;(2)"有目的试误"策略不仅有助于解答"交通"一题,其在体现个体问题解决能力中仍处于重要位置;(3)依据使用"有目的试误"策略的程度,学生的问题解决过程可分为5个群组(最优组、有目的试误组、其他策略组、只顾玩乐组、功能障碍组),台湾在"最优组"表现最好,香港在"只顾玩乐组""功能障碍组"比例最高。可见,log数据在分析问题解决过程、辨别问题解决群组、发展基于证据的教育决策等方面有其重要作用。

Log data is a subset of educational big data, and it provides a new way to analyze problemsolving process. This paper captures the log data of China in PISA（2012）, and adopts educational data mining methods such as relationship mining, clustering to analyze students＇ problem-solving time, mouse clicks, and the use of ＂purposeful trial-and-error＂ when dealing with the item TRAFFIC. The results indicate that there is a sharp contrast between students in Taiwan and those in Shanghai, while students in Shanghai have lower problem-solving time and mouse clicks suggesting that they may be more concerned about time cost.（2） The ＂purposeful trial and error＂ strategy not only helps students to solve the item TRAFFIC, but is still in an important position in the embodiment of individual problem-solving ability.According to the degree of using this strategy, problem-solving process can be divided into five groups： the optimal group, the purposeful trial-and-error group, other policy group, the play group, and the dysfunction group. Taiwan performes best in the optimal group, while Hong Kong has the highest ratio of ＂play group＂and ＂dysfunction group＂. Thus, log data plays an important role in analyzing problem-solving process,identifying problem-solving groups, and developing evidence-based educational decision-making.