Palm Oil Production Prediction Using Support Vector Regression Algorithm and Long Short-Term Memory

Currently, palm oil plantations play an essential part of the agricultural sector, especially in the worldwide palm oil supply network. The global expansion of palm oil plantations has been swift, and Indonesia and Malaysia are expected to maintain their dominance in the export of vegetable oil. Palm oil produced by one company, which has 20 plantations spread across Riau, sometimes experiences fluctuations, both increases and decreases, in the previous period. This often occurs throughout the period, with a significant decline in production. This trend of fluctuations has raised concerns among parties facing uncertainty and risk in palm oil trading, and it affects the income of small farmers, which in turn impacts national revenue in the long term. An effective approach needs to be taken by predicting the production volume based on data from a specific period. Many techniques can be used for prediction, as has been done in previous research. However, this study applies a more consistent technique by using Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) models. As per the research findings, the SVR model outperforms the LSTM model, as evidenced by the SVR’s error value of $0.0 \%$. This highlights the SVR model’s superior performance compared to the LSTM model. Based on the SVR model’s performance reaching $100 \%$, this model can be used as a reference for predicting the production quantity of other types such as sunflower oil, olive oil, corn oil, or even rubber, tea, and similar products using time series data.