<![CDATA[Service in the world of education is an important element for the creation of an academic atmosphere that is conducive to the implementation of a successful teaching and learning process. The process of service to students there is a tendency to be implemented not following the minimum service standards that must be provided to students so that students tend to complain about the services provided. Submission of criticism, complaints, input, or suggestions for dissatisfaction and problems that exist in the university environment is still very limited. Complaints can be constructive if submitted to the right place and party. In this research the data processing of email complaints from students conducted at the academic student body (students.bsi.ac.id). Student complaint data that will be processed is data in the form of * .xls complaint file. Before text data is analyzed using text mining methods, the pre-processing text needs to be done including tokenizing, case folding, stopwords, and stemming. After pre-processing, the classification method is then performed in classifying each complaint category and dividing the status into two parts, namely complaint and not complaint so that the status becomes a normal condition in text mining research. The purpose of this study is to obtain the most accurate algorithm in the classification of student complaints and can find out the results of the classification of the Naïve Bayes algorithm method and Support vector Machine used and compared. In this study, the results of testing by measuring the performance of these two algorithms using Cross-Validation, Confusion Matrix, and ROC Curves. The obtained Support vector Machine algorithm has the highest accuracy value compared to Naïve Bayes. AUC value = 0.922. for the Support vector machine method using the student academic data collection dataset (students.bsi.ac.id) has 84.45%, from the Naïve Bayes algorithm has an accuracy rate of about 69.75% and AUC value = 0.679.]]>
