In order to improve the accuracy of short-time traffic flow prediction, an improved LSSVM-based short-time traffic flow prediction model is proposed. To address the problem that the traditional hybrid frog-jumping algorithm (SFLA) easily falls into local optimum, an improved hybrid frog-jumping algorithm (ISFLA) based on a new local update strategy is proposed, which is combined with the least squares support vector machine (LSSVM) to improve the prediction capability of LSSVM by using this algorithm to optimize the key parameters of LSSVM. The model and algorithm are simulated and analyzed with examples to prove the feasibility of the model and the effectiveness of the algorithm.