Throughout the 20th century, terrorism has significantly impeded the stable development of the international community. Evaluating the potential consequences of terrorist attacks is crucial to curb terrorist activities, and the deployment of counter-terrorism resources is of significant importance. In this paper, we propose an algorithm that combines a processing strategy of multiple types of features with the Light Gradient Boosting Machine (LightGBM) model to predict the severity of consequences that may result from terrorist attacks. Our approach utilizes RF-RFE, a random forest-based recursive feature elimination method, to select important features. Furthermore, we integrate LightGBM and text representation techniques to merge information from different feature types and apply genetic algorithms to fine-tune the LightGBM model. We evaluate the effectiveness of our approach using the Global Terrorism Database (GTD), achieving AUC and accuracy rates of 91.03% and 73.93%, respectively. Our research leverages machine learning to predict various consequences of potential terrorist attacks, thereby assisting decision-makers in formulating specific measures tailored to distinct scenarios.