Resistors play an indispensable role in modern electronic devices. The variation of resistor values is often nonlinear and influenced by multiple factors, and changes in resistor values directly affect the performance of electronic components. Accurately predicting resistor value changes has thus become a research focus. To address this, we propose a combined model based on Optuna-VMD-Random Forest to predict resistor values. VMD decomposes the signal to remove noise, providing high-quality data input for the model; Optuna is responsible for optimizing the model to find the best configuration parameters; and Random Forest is used for predicting resistor values, leveraging its ensemble characteristics to handle the nonlinearity and noise resistance in resistor value prediction. Compared with a standalone Random Forest model, the proposed combined model improves MAE, RMSE, and MAPE by 16.6%, 26.3%, and 15.1% respectively, significantly enhancing prediction accuracy.