Loan Default Prediction - Imperial College London 贷款默认预测-伦敦帝国理工学院

您用此数据集确定贷款是否会违约，以及如果违约所造成的损失。与传统的基于财务的方法来解决此问题不同，在传统方法中，我们以二进制方式区分好与坏交易对手，我们力求预见并合并所造成损失的违约和严重性。为此，我们正在建立传统银行业务和资产管理之间的桥梁，在传统银行业务中我们正在寻求减少经济资本的消耗，在资产管理方面我们可以优化金融投资者的风险。数据集是由伦敦帝国学院的研究人员提供。

This dataset is used to determine whether a loan will default and the losses incurred if a default occurs. Unlike traditional finance-based approaches to this problem, which binary-classify counterparties as either "good" or "bad", our work seeks to anticipate defaults and the severity of losses arising from such defaults, and aggregate the total resulting losses. To this end, we aim to bridge traditional commercial banking and asset management: in traditional banking, we seek to reduce economic capital consumption, while in asset management, we can optimize the risk profiles of financial investors. This dataset was provided by researchers at Imperial College London.

搜集汇总

背景与挑战

背景概述

该数据集由伦敦帝国学院提供，用于预测贷款违约及其导致的损失，不仅区分违约与否，还关注损失程度。数据包含超过20万条独立观测和近800个特征，涉及标准化、去趋势化的金融交易信息，要求预测测试集中0到100范围内的损失值。