Genetic Algorithm-Based Neural Network for Vegetable Price Forecasting on E-Commerce Platform: A Case Study in Malaysia

With the rapid growth of E-Commerce platforms, accurate price forecasting is increasingly vital in Malaysia’s agricultural sector. This sector faces various challenges, including weather uncertainties, supply–demand imbalances, market inefficiencies, fluctuating input costs, and global market influences. To address these challenges, a vegetable price forecasting model is proposed, integrating data and trends to enable informed decision-making. Neural networks have shown promise in capturing complex patterns, but overfitting remains an obstacle, limiting their performance on unseen data. To overcome this, the study introduces a genetic algorithm-based neural network (GANN) approach for vegetable price forecasting. GANN combines a genetic algorithm with a recurrent neural network and focuses on forecasting spinach (bayam), round wax gourd (kundur bulat), galangal (lengkuas), and holland potatoes (ubi kentang holland) prices. The historical monthly vegetable price data from 2010 to 2021, which are obtained from the Federal Agricultural Marketing Authority (FAMA) Malaysia, are utilized in doing the forecasting. Comparative analysis with the multilayer perceptron (MLP) technique reveals that GANN outperforms MLP in terms of forecasting accuracy. This study contributes to the advancement of price forecasting methods for vegetables on E-Commerce platforms, providing valuable insights for businesses.