In order to fully examine how driving behavior affects vehicle energy consumption and further cut down on excessive energy use while driving, an energy-saving evaluation method for driving behavior based on the SHAP (Shapley Additive exPlanations) scorecard model is proposed. First, to categorize short trip segments, a LightGBM model is built, which performs better and has high interpretability. The SHAP method is then used to statistically assess the categorization findings in order to ascertain how various driving events and driving behavior characteristics affect energy consumption. After that, the data are converted into scores using a scorecard model, which is then used to examine the evaluation outcomes of each sample and individual driver. Lastly, by checking the distribution of scores and the pertinent factors for various scores, the scorecard model’s correctness and reasonableness in evaluating energy-saving driving practices of short-trip segments are confirmed.