This paper presents a novel method for efficient state of charge (SoC) estimation in electric vehicle (EV) batteries using the Extra Tree Regressor (ETR) and Light Gradient Boosting (LightGBM) algorithms. The study aims to improve the accuracy and reliability of SoC estimation, which is crucial for optimizing EV performance and energy utilization. The research involves a comprehensive dataset derived from various driving cycles and battery records, ensuring the model's applicability to different driving patterns and conditions. The ETR and LightGBM models are chosen for their robustness and ability to handle complex, high-dimensional datasets. The ETR model, with its ensemble learning approach, effectively handles noise and overfitting, while LightGBM excels in speed and prediction accuracy. The experimental results demonstrate that the ETR model outperforms LightGBM in terms of higher R² values (0.9983), lower Root Mean Square Error (RMSE) (0.62), Mean Absolute Error (MAE) (0.085), and Mean Squared Error (MSE) (0.39). The study highlights the significance of battery capacity in accurate SoC estimation and provides a robust tool for improving energy management in EVs. The proposed method is validated through rigorous testing and comparison with other models, demonstrating its superior performance in SoC prediction for EV batteries.This paper presents a novel method for efficient state of charge (SoC) estimation in electric vehicle (EV) batteries using the Extra Tree Regressor (ETR) and Light Gradient Boosting (LightGBM) algorithms. The study aims to improve the accuracy and reliability of SoC estimation, which is crucial for optimizing EV performance and energy utilization. The research involves a comprehensive dataset derived from various driving cycles and battery records, ensuring the model's applicability to different driving patterns and conditions. The ETR and LightGBM models are chosen for their robustness and ability to handle complex, high-dimensional datasets. The ETR model, with its ensemble learning approach, effectively handles noise and overfitting, while LightGBM excels in speed and prediction accuracy. The experimental results demonstrate that the ETR model outperforms LightGBM in terms of higher R² values (0.9983), lower Root Mean Square Error (RMSE) (0.62), Mean Absolute Error (MAE) (0.085), and Mean Squared Error (MSE) (0.39). The study highlights the significance of battery capacity in accurate SoC estimation and provides a robust tool for improving energy management in EVs. The proposed method is validated through rigorous testing and comparison with other models, demonstrating its superior performance in SoC prediction for EV batteries.