车辆贷款违约预测（讯飞A.I算法赛）
收藏资源简介：
随着监管政策步入关键落地期，受合规监管标的限额影响，曾备受追捧的大额标的逐渐消失，小额分散的车贷业务成为网贷平台转型的主要方向之一。车贷资产由于进入门槛低、借款额度低、流动性高、限期短等优点，但做好风险防控依然是行业的主要问题之一。 国内某贷款机构就面临了这样的难题，该机构的借款人往往拖欠还款或拒不还款，导致该机构的不良贷款率居高不下。面对如此头疼的问题，该机构将部分贷款数据开放，诚邀大家帮助他们建立风险识别模型来预测可能违约的借款人（敏感信息已脱敏）. 给定某机构实际业务中的相关借款人信息，包含53个与客户相关的字段，其中loan_default字段表明借款人是否会拖欠付款。任务目标是通过训练集训练模型，来预测测试集中loan_default字段的具体值，即借款人是否会拖欠付款，以此为依据，降低贷款风险。
As regulatory policies enter their critical implementation phase, affected by the loan amount limits imposed by compliance supervision, once highly sought-after large-value loan projects have gradually disappeared, and small-amount, diversified auto loan businesses have become one of the main directions for online lending platforms to undergo transformation. Auto loan assets boast advantages such as low entry barriers, low loan amounts, high liquidity, and short repayment tenors, yet effective risk prevention and control remains one of the core challenges facing the industry. A domestic lending institution is grappling with exactly this predicament: its borrowers frequently default on or refuse to repay their loans, resulting in a stubbornly elevated non-performing loan (NPL) ratio for the institution. To tackle this pressing problem, the institution has made partial loan data available, inviting participants to develop risk identification models for predicting potentially defaulting borrowers (sensitive information has been anonymized). Relevant borrower information from the institution's actual business operations is provided, which contains 53 customer-related features. The loan_default field indicates whether a borrower will default on their payment obligations. The task objective is to train a model on the training dataset to predict the specific values of the loan_default field in the test dataset—namely, whether a borrower will default on their payments—thereby providing a basis for reducing lending risks.
