Pattern Analysis To Prediction of Graduation With CPAR Algorithm

Abstract

This paper describes the steps to find the connection between subject on academic achievement. The purpose of this paper for it to see the condition of students. Steps undertaken include domain understanding, collect data, preprocessing, selection of data mining, data processing, evaluation and presentation patterns. Prepocessing includes data cleaning, data integration, data selection and data transformation. In one of the data cleaning is done the equation GPA for students who drop out. Integration done by combining all the data. The selection is done by determining the data attributes that will dianalis. And data transformation is done by doing some math operations. This paper also describes how the student graduation prediction using CPAR algorithm. The data is divided into the positive data and negative data. The next step is to calculate the TWT, forming Gain, and calculate the Laplace accurate and establish rules based on the variation of the gain similarity ratio. The benefits of this research is that we can predict gradution rates students so that early anticipation can be done as aprogram of counseling, special additional duty, or other similar programs. Keywords : data mining, CPAR, algorithm, warehouse