Data Mining for Instructional Design, Learning and Assessment

Abstract

Statistical analysis is widely used in many different areas: medicine, business, natural and social sciences, and of course, in education. In this last topic, it is common that teachers make simple statistical analysis on the results of the students at the end of an exam or a course, and this is useful for the evaluation of that course. However a more powerful use of statistics can and must be done if the analyses are used to modify the methodology of learning personalizing contents and methods for groups of students with similar skills. To make a realistic personalization of learning, data mining techniques must be used. They are also useful to manage big amounts of information mainly composed by: contents, skills, tools, grades and students. In this chapter, we present data mining techniques used in instructional design, in learning and in the assessment of the students. In order to reduce, interpret and classify the information, factor and cluster analysis have been used. Factor analysis is a technique that extracts few unobserved new variables (factors) from a big number of data. These factors are linear combinations of the observed variables and the expert analyzer must define the information that underlies each factor. Cluster analysis classifies all the information in some sets (clusters) of items with common features. Let's present here two examples of the use of Data Mining in e-learning:  Example 1. An institution must decide its learning methodology, and it has planned to use a Learning Management System (LMS). Of course, an LMS contains many tools, and teachers and students must learn how to use these tools. But not all these tools add value to learning, and probably many of them are redundant, that is, students can acquire the same competences using different tools. In (Vicent, 2007) teachers were asked to value (from 0 to 3) the performance of each tool (24 were considered) to develop each skill. Using factor and cluster analysis, an LMS of only 5 tools was defined to run an engineering online degree in the European Higher Education Area.  Example 2. If an LMS is used for learning, much information of the students is available: results of questionnaires, number of post in the forums, number of visits to the contents, etc. It is possible to classify the students in function of their behavior with a cluster analysis. This way, lazy, willing, active, brilliant, etc. 8 Data Mining for Instructional Design, Learning and Assessment 147