Mobile learning for assessment
Title: Pattern capacity participants exam of mobile learning for assessment
Authors: Heri Nurdiyanto, Herman Dwi Surjono, Priyanto
Publication: International Journal of Scientific and Technology Research 8 (9), pp.713 indexed by Scopus.
Abstract:
The adaptive assessment system is expressed as an interactive approach to assessing the learner in the learning system. Stages undertaken in the development of this system include determination of bank questions, determination of the initial ability level of examinees, selection of items, assessments, termination of tests, and conclusions about the ability of examinees. Determining the initial ability level of examinees is very important because its accuracy dramatically affects the effectiveness of a selection of questions. Rule-based methods are used to extract information, rule-based methods combined with machine learning techniques are proposed to assess the level of ability of regular students and students with special needs. Machine learning techniques used are Naive Bayes, Multilayer Perceptron, SMO, Decision Tree, JRIP, and J48. The best accuracy results are achieved using the JRIP rule-based method of 64.12. The rules for the determination of the level of ability are formed based on expert opinion. The strength of examinees to vary and the amount of data evolving lies in need for dynamic formation of rules. The discovery of patterns in the test data of the participants can be used as the basis for the creation of states to replace the expert as well as improve the prediction accuracy. It is necessary to extract the pattern so that it can be used for the formation of the initial capability rules for examinees.
