Abstract: The Education Data Mining (EDM) is the data mining for analyzing the data from and for education. This paper proposed the framework to analysis the undergraduate program. The selected data mining technique is the well-known Decision Tree Data Mining Technique. This result of this proposed is able to analysis the major subjects that will effect with the student and the registered planning. Moreover, the course management is able to plan the relationship and the program. The experimental result tested from computer simulation from real registration data and grade of department of Informatics, Mahasarakham University shows that the knowledge representation of the program analysis using Decision Tree Data Mining Technique is very well. The average accuracy from k-fold cross validation is 74.05%.

• Publication Date: 27-Aug-2014
• DOI: 10.15224/978-1-63248-024-8-18
• Views: 0
• Downloads: 0

Abstract: We introduce a truss decomposition algorithm for bipartite graphs. A subgraph G of a graph is called k-truss if there are at least k-2 triangles containing any edge e of G. By a standard breadth-first-search algorithm, we can compute the truss decomposition for large graphs. To extract a dense substructure that represents community in graph G, this method avoids the intractable problem of clique detection. The truss decomposition is not, however, applicable to the bipartite graphs due to its definition. For this problem, we have proposed quasi-truss decomposition introducing an additional set of edges. For this decomposition, there is another problem such that dense subgraphs G1 and G2 are connected with a small number of edges. The previous algorithm detects the sparse structure H = G1 ⋃ G2 as quasi-truss due to the definition. In this paper, we improve the algorithm to extract denser substructures by removing such sparse edges with a top-down strategy. The extended algorithm has been

• Publication Date: 27-Aug-2014
• DOI: 10.15224/978-1-63248-024-8-19
• Views: 0
• Downloads: 0