Accurately estimating the threat value of air targets has important reference significance for air defense combat command decision-making. In view of the complex features of aerial targets that easily cause model over-fitting and the sine-cosine algorithm is prone to premature and fall into local optimality, this paper uses the Least Absolute Shrinkage and Selection Operator (LASSO) to remove the redundant features of the target, and then improves the sine cosine algorithm (SCA) with some strategies such as good point set initialization population, nonlinear amplitude adjustment factor, random inertia weight, adaptive end point weight, and uses the improved Sine Cosine algorithm to optimize the support vector Regression (SVR) model, and a target threat estimation model based on lasso algorithm and improved sine cosine optimized support vector regression is constructed. The comparative experimental results show that the improved sine and cosine algorithm enhances the global search ability and local convergence speed, and the obtained target threat estimation model has high accuracy and stability, which can provide a scientific reference for air defense combat command decision-making.