Investigation of student success at faculty of engineering by using data mining

Cluster analysis, using the different characteristics or similar properties of objects in the data set, aims at creating in the same cluster homogeneous and between different clusters heterogeneous groups. In other words, the objects that make up a cluster that is so similar to each other and discrete clusters in the clustering process, and different from each other as much as it has been so successful. In this thesis, using the methods of data mining operations carried out analysis of the student. This is the process of analyzing students' demographic data, and settlement in University Entrance Exam scores success percentages weighted grade point average information gained will be used. A clustering process was carried out according to school types and students regions. In addition, a particular section of the students were surveyed. Through these surveys examined how it affects the success of the students' family information. Clustering methods, fuzzy clustering and hard clustering to be discussed under two main headings in comparison. In this context fuzzy clustering methods, fuzzy C-means, Gustafson-Kessel, and Gath-Geva algorithms and hard clustering methods, k-means and k-medoids algorithms are explained in detail, and an assessment of the achievements made by applying on data set. Cluster validity and Box-Plot analysis methods were used in determining the success status.