【摘要】 现阶段，二手船估值易受主观判断的影响，且存在估值时未全面考虑影响价格的因素等问题。为此，综合考虑船舶自身因素和市场因素，采集二手散货船在2000—2022年的交易数据进行实验，采用决策树、随机森林、XGBoost、LightGBM等4种机器学习算法分别建立二手散货船估值模型。为进一步提高估值的准确性，将后3种模型进行加权平均模型融合。结果显示：单个模型在测试集上的最高决定系数值为0.879,而融合模型的决定系数值为0.889。这表明所提出的融合模型能较为准确地对二手散货船进行估值。所提出的融合模型可以为航运等相关行业提供更准确、客观的二手散货船估值服务。更多还原

【Abstract】 At present, the valuation of second-hand ships is susceptible to subjective judgment, and there is incomplete consideration of price-influencing factors in the valuation. To address these problems, the self-factors of ships and the market factors are considered comprehensively, transaction data of second-hand bulk carriers from 2000 to 2022 are collected for experiments, and four machine learning algorithms, decision tree, random forest, XGBoost, and LightGBM, are employed to construct valuation models for second-hand bulk carriers, respectively. To further enhance the valuation accuracy, the latter three models are fused as a weighted average model. The results demonstrate that, the highest coefficient value of determination for a single modle on the test set is 0.879, while the coefficient value of determination for the fusion model is 0.889. This indicates the proposed fusion model can accurately value second-hand bulk carriers. The proposed fusion model can provide more accurate and objective valuation services for second-hand bulk carriers in shipping and related industries.更多还原