Autonomous driving, particularly when it involves smart decision-making and path planning in dynamic settings such as highways, presents far greater challenges compared to navigating static environments. This research paper investigates the effectiveness of various machine learning models in predicting the trajectories of surrounding vehicles, focusing on the personalized framework using LSTM-XGboost model. In this research, we have categorized the vehicles NGSIM dataset into Traditional, Moderate, and Aggressive driving styles to assess model performance using RMSE and MAE metrics across different prediction horizons (1- 5 seconds). The results demonstrate that the LSTM-XGboost model consistently outperforms other baseline models, including GNN, LSTM, GBM-LSTM, and ARIMA, in all vehicle categories. Notably, the personalized LSTM-XGboost model, tailored to specific vehicle categories, yields significantly better prediction accuracy compared to the overall dataset. This research highlights the potential of advanced machine learning models for improving vehicle trajectory predictions and underscores the importance of vehicle categorization for enhancing model performance.