基于随机森林预测彩礼返还比例

Predicting the Return Ratio of Bride Price Based on Random Forest

摘要: 彩礼是我国民间婚俗的重要组成部分，近年来因婚约解除引发的返还彩礼纠纷日益受到关注，但基于真实裁判文书数据的量化研究仍较为缺乏。本文从中国裁判文书网获取湖南省2024年涉及彩礼返还问题的104份民事判决书，系统提取了双方是否领证、是否举办婚礼、是否同居、共同生活时间、子女生育情况、各项彩礼及相关财物金额等关键事实特征。首先采用K均值聚类分析当事人群体类型，进而构建随机森林回归模型对彩礼正礼、三金五金及红包礼金的返还比例进行预测与对比。结果表明，彩礼正礼金额、子女生育个数、共同生活时间是影响返还比例的最重要因素；而是否领证、是否举办婚礼的影响相对微弱。本文为彩礼返还纠纷的司法裁判提供了数据驱动的参考，也为当事人理性预估诉讼结果提供了实证依据。

Abstract: Bride price is an important component of traditional Chinese marriage customs. In recent years, disputes over the restitution of bride price arising from the dissolution of engagements have attracted growing public attention, yet quantitative studies based on authentic court judgment data remain scarce. This paper retrieves 104 civil judgments from Hunan Province in 2024 that involve bride price restitution issues from the China Judgments Online website, and systematically extracts key factual characteristics, including whether the parties obtained a marriage certificate, whether a wedding ceremony was held, whether they cohabited, the duration of cohabitation, the number of children born, and the amounts of various bride-price items and related properties. First, K-means clustering is used to analyze the types of party groups; then a random forest regression model is constructed to predict and compare the restitution ratios of the core bride price, gold jewelry (customarily referred to as the “three golds” and “five golds”), and red-envelope cash gifts. The results indicate that the amount of the core bride price, the number of children, and the cohabitation duration are the most influential factors affecting the restitution ratio, whereas the effects of whether a marriage certificate was obtained and whether a wedding ceremony was held are relatively minor. This study provides a data-driven reference for judicial adjudication in bride-price-restitution disputes and offers empirical evidence for the parties to rationally estimate possible litigation outcomes.