Classification analysis is a method used to classify or analyze the relationship between several predictor variables and response variables that aim to predict the class of an object whose label is unknown. This classification problem arises when a number of measures consist of one or more categories that cannot be defined directly but use a measure. MARS is one of the classification methods focused on overcoming high-dimensionality and discontinuity problems in data. The accuracy or classification level of the MARS method can be improved using a resampling method, namely bagging. This study will apply the MARS model to obtain a model for classifying the status of people with diabetes based on people with diabetes. The data used in this study is secondary data obtained from the Kaggle website which can be accessed through https://www.kaggle.com/uciml/pima-indians-diabetes-database, namely the Pima Indians Diabetes Database and processed using R software. The results of MARS modeling concluded that the probability of someone having diabetes is 0. The probability of someone not having diabetes is 1, with a classification accuracy of 81.38%. In contrast, the accuracy of the best MARS bagging method among 200 replications is 75.23%, so in this study, a more appropriate method is used to classify the status of people with diabetes.