PREDIKSI NILAI EKSPOR KOMODITAS PERKEBUNAN DI PASAR INTERNASIONAL DENGAN ALGORITMA LINEAR REGRESION

The development of Indonesia's agricultural sector, particularly plantation crops, has made rapid progress thanks to abundant natural resources and a supportive climate. This sector is divided into perennial and seasonal crops, making a significant contribution to the national economy. High international demand drives further expansion opportunities, although it still faces competitiveness challenges. The application of predictive technology, such as Linear Regression algorithms, can help project future market prices, enabling more accurate decision-making. This study shows that the mean imputation method for handling missing values is effective in predictive analysis, with an MSE of 0.3 and a MAPE of 14.4% for coffee commodities in Japan. The study used seven years of data from 2013 to 2023, covering 29 commodities. Various imputation methods, such as mean, median, mode, spline interpolation, and linear interpolation, were tested to handle missing values. Among these methods, mean imputation proved most effective in addressing data gaps. Additionally, this study highlights that the use of Linear Regression algorithms not only enhances efficiency in data analysis but also maximizes profitability. With this technology, farmers and industry players can make more precise decisions based on accurate price projections, thereby optimizing their yields and incomes. Implementing Linear Regression improves efficiency and profitability, supports the stability of plantation crop exports in international markets, and ultimately contributes to sustainable national economic growth.

Keywords: Price Prediction, Agricultural Commodities, Missing Value, Imputation, Normalization, Linear Regression, Streamlit.