No abstract

Higher education has changed a lot in the last decade. The use of various innovative techniques and technologies, especially ICT in teaching learning process is increasing day by day. Number of information systems has been developed and successfully implemented to support educational processes. These systems typically capture almost every data regarding a student, right from their enrollment into a course to graduation and placement. If these data are analyzed and visualize properly, can provide valuable knowledge that can be used to enhance their learning skill and to predict threats if any, well in advance so that appropriate measure can be taken to avoid it. Knowledge Discovery in Database (KDD) is usually referred as Data mining. It is a process of extracting new and potential useful information from large databases. Data mining tools are used to identify any pattern or predict future trends and behaviors. This enables decision maker to make proactive and knowledge-driven decisions. This paper presents an application of data mining in higher education. Association rule mining is applied to analyze the performance of students in their examinations and predicts the outcome of the forthcoming examination. This prediction allows student and teacher to identify the subjects which need more attention even before the commencement of semester.

No abstract

Problem statement: Classification techniques play an important role in Data Mining. Large number of classification techniques has been proposed in the literature. No single algorithm can be considered optimal for all type of data set. Accuracy of classification result highly depends on the selection of classification algorithms. Different classification techniques produce different results for the same data set. Thus finding the optimal algorithm for the given data set is a challenge. The outcome of this research work can be useful in selecting most suitable classifier for the given dataset. Research Methodology: To determine the effectiveness of various classification algorithms, authors run some well-known classification algorithms against some standard datasets. Effectiveness of various algorithms is measured on the basis of average accuracy, time taken to build classification model, mean absolute erroretc. Results: Based on the comparative study of the experiment results, authors suggest the optimal algorithm for different categories of datasets.

No abstract