Classification and prediction are some of the capabilities of Data Mining. This study will implement a classification model using the Fisher Linear Discriminant (FLD) function. After the classification model is obtained, the model is used to predict the Grade Point Average category (GPA-1st). The FLD classification models used are 9 models derived from cumulative student data from 2008 to 2016 academic year. In the FLD model, GPA-1st is used as the dependent variable, while the factors of high school location, high school status, high school type, and English proficiency level are used as independent variables. These models are used to predict the GPA-1st category for students in 2017. Crosstab tables are used to measure the accuracy of the classification model and accuracy of the... prediction model. As the result, the accuracy average of the 9 classification models in students' GPA-1st is 68.67%. While the accuracy average of predictions using 9 models is 58.28%.