Instance-based learning has been proved to be effective for credit assessment in Peer-to-peer(P2P) lending. A key challenge of this application is how to measure the similarity of loans, which have usually multiple features gained from different data sources and models. In this paper, a new similarity measure is proposed to effectively integrate the information from different sources and models for credit assessment in P2P lending. Specifically, we firstly deconstruct the characteristics of P2P lending and present four heterogeneous distance functions to measure the loans’ similarity, which are generated by different models and information sources. Then, we propose an integrated similarity measure that combines the above similarities by minimizing their conflicts, which overcomes the bias of the single model and single information source. Finally, we employ the portfolio selection model to develop our investment strategy. Experimental results using real datasets from Prosper demonstrate that our integrated similarity measure improves the performance of the instance-based credit assessment in P2P lending.

基于实例的学习已被证明对P2P借贷中的信用评估有效。该应用程序的一个关键挑战是如何衡量贷款的相似性，这些相似性通常具有从不同数据源和模型获得的多个功能。在本文中，提出了一种新的相似性度量，以有效地整合来自不同来源和模型的信息，以进行P2P借贷中的信用评估。具体来说，我们首先解构P2P贷款的特征，并提出四个异类距离函数来衡量贷款的相似性，这是由不同的模型和信息源生成的。然后，我们提出了一种综合相似性度量，该度量通过最小化它们之间的冲突来组合上述相似性，从而克服了单一模型和单一信息源的偏见。最后，我们采用投资组合选择模型来制定我们的投资策略。使用来自Prosper的真实数据集的实验结果表明，我们的集成相似性度量提高了P2P借贷中基于实例的信用评估的性能。