Educational Data Mining: Analyzing Teacher Behavior based Student's Performance

The useful information is an expensive and import resource for decision support. The getting of it empowering the institution decisions. While data mining is emerging field that can provide the techniques, producing the useful information from the raw data that available. However educational data mining can provide this information cheaply through mining huge data that are produced from different educational activities in academic institutions. More descriptive and predictive information with different aspects can be obtained; behavior of examiner in evaluating the student performance is one of them and has important role in analyzing the course and its lecture. This paper was conducted in this field, it aims to study the performance of students according to the teacher also to the course level in order to detect abnormal behavior of teacher in evaluating his students. The materials would be used in this work is clustering task; one of the data mining. It was used as a tool for describing his behavior by grouping individuals. The data that was supplied from case study, was constructed in simulation from Iraqi's university system, degrees of students in each course, then was transformed to statistical features such as minimum, maximum, standard deviation. So each course was identified by these feature and passed to kmeans algorithm for clustering. The results reveal significant bias to behavior teacher in evaluating the student, however the features of course for specific teacher grouped in most of time for the same cluster, in the same time there is bias to the level and semester that course belong. This work can be extended with large data and more courses that enable for better results. Also there is a chance for more analysis in different aspects of academic interest fields and another data mining tasks with different techniques.