ANALISIS PERBANDINGAN METODE TRADISIONAL DAN MACHINE LEARNING PADA PERAMALAN HARGA BAWANG MERAH DI NUSA TENGGARA BARAT

Abstrak

The price of red onions often fluctuates, which can negatively impact economic stability and potentially lead to inflation. Therefore, accurate predicting of red onion prices is necessary. This study analyzes the comparison between Traditional Methods and Machine Learning in predicting red onion prices. The objective is to evaluate the effectiveness of various prediction methods based on the Mean Absolute Percentage Error (MAPE) and to determine the best method for predicting red onion prices. In the statistical approach, three methods were used: Double Moving Average with a MAPE value of 21.954, Double Exponential Smoothing with a MAPE value of 15.1163, and Autoregressive Integrated Moving Average with a MAPE value of 26.31141. In the Machine Learning approach, two methods were used: Support Vector Machine with a MAPE value of 1.88, and Artificial Neural Network with a MAPE value of 29.2249. Based on the comparison of MAPE values, the Support Vector Machine method from Machine Learning is the most effective for predicting red onion prices in West Nusa Tenggara.