ANALISIS PERBANDINGAN METODE REGRESI LINIER DAN EXPONENTIAL SMOOTHING DALAM PERAMALAN PERMINTAAN PERTALITE PADA PT PERTAMINA PATRA NIAGA REGIONAL SUMATERA BAGIAN UTARA (SALES AREA RETAIL MEDAN)

Abstract

This research discusses the comparative analysis of linear regression and exponential smoothing methods in forecasting demand for Pertalite products at PT Pertamina Patra Niaga Regional North Sumatra, especially the Medan Retail Sales Area. This research aims to find out the most accurate forecasting method in predicting demand for Pertalite in the Medan Retail Sales Area. Historical demand data from the period January 2023 to June 2024 is used as the basis for the research. The research results show that the linear regression method has a better level of forecasting accuracy than the exponential smoothing method based on error calculations. The linear regression method produces a Mean Absolute Deviation (MAD) value of 3,157,216, a Mean Square Error (MSE) of 1,571,886,522, and a Mean Absolute Percent Error (MAPE) of 2.97%. Meanwhile, the exponential smoothing method with alpha values of 0.1, 0.5, and 0.9 produces MAD values of 3,244,072, 3,746,722 and 4,796,245 respectively, MSE values of 2,010,507,177, 2,580,704,082 and 3,344,564,362, and MAPE values of 3.08%, 3.54% and 4.52%. Therefore, it is concluded that the linear regression method is the most appropriate to use in forecasting demand for Pertalite in the Medan Retail Sales Area because it has the smallest error value. It is hoped that this research can help PT Pertamina Patra Niaga in making strategic decisions regarding inventory management and fuel distribution more efficiently.