Classifiers for educational data mining

The idea of classification is to place an object into one class or category, based on its other characteristics. In education, teachers and instructors are all the time classifying their students for their knowledge, motivation, and behaviour. Assessing exam answers is also a classification task, where a mark is determined according to certain evaluation criteria. Automatic classification is an inevitable part of intelligent tutoring systems and adaptive learning environments. Before the system can select any adaptation action like selecting tasks, learning material, or advice, it should first classify the learner’s current situation. For this purpose, we need a classifier – a model, which predicts the class value from other explanatory attributes. For example, one can derive the student’s motivation level from her/his actions in the tutoring system or predict the students who are likely to fail or drop out from their task scores. Such predictions are equally useful in the traditional teaching, but computerized learning systems often serve larger classes and collect more data for deriving classifiers. Classifiers can be designed manually, based on expert’s knowledge, but nowadays it is more common to learn them from real data. The basic idea is the following: First, we have to choose the classification method, like decision