The aim of this research is to examine and forecast the sales volume of Aquviva beverages at UD Arkamart by applying Support Vector Regression (SVR) with a Radial Basis Function (RBF) kernel. The study follows a quantitative approach and relies on daily sales records collected directly for the period of June 1 to June 30, 2025. The system was built with the Rapid Application Development (RAD) approach, a software development methodology that prioritizes speed, flexibility, and ongoing iteration through quick prototyping. When the model was evaluated on the whole dataset, it produced a Mean Absolute Error (MAE) of 1.035, a Mean Squared Error (MSE) of 1.229, and a Root Mean Squared Error (RMSE) of 1.109. A Mean Absolute Percentage Error (MAPE) of 16.42% places the prediction error in the “good forecasting” category, which shows that the model can track the pattern of daily sales fluctuations. Hence, SVR can be regarded as a well-performing method and a suitable basis for estimating inventory needs. Keywords: Support Vector Regression; Rapid Application; Radial Basis Function; Prototipe Abstrak Studi ini bertujuan menganalisis sekaligus memprediksi volume penjualan minuman Aquviva di UD Arkamart dengan metode Support Vector Regression (SVR) yang memakai kernel Radial Basis Function (RBF). Pendekatan yang digunakan bersifat kuantitatif, dengan data yang diperoleh langsung dari catatan penjualan harian pada 1 hingga 30 Juni 2025. Sistem dibangun melalui metode Rapid Application Development (RAD), yakni pendekatan pengembangan perangkat lunak yang mengutamakan kecepatan, fleksibilitas, dan perbaikan berulang lewat purwarupa yang dibuat dengan cepat. Pengujian model pada seluruh data menghasilkan Mean Absolute Error (MAE) sebesar 1,035, Mean Squared Error (MSE) sebesar 1,229, dan Root Mean Squared Error (RMSE) sebesar 1,109. Mean Absolute Percentage Error (MAPE) sebesar 16,42% menandakan bahwa kesalahan prediksi masuk kategori baik (good forecasting), sehingga model dapat mengikuti pola naik turun penjualan harian. Dengan demikian, metode SVR berkinerja baik dan layak dijadikan landasan untuk memprediksi stok barang.
Copyrights © 2026