基于GBDT算法的大数据风控模型研究

Research on Big Data Risk Control Model Based on GBDT Algorithm

大数据风控是指金融机构利用机器学习的方法,以行为大数据替代抵押资产,进行信用风险评估,从而解决长尾客户风控难问题。GBDT算法是一种基于梯度提升的高效集成学习算法,本文基于GBDT算法建立大数据风控模型,并针对LendingClub的个人信贷真实数据进行实证研究,结果表明基于GBDT算法的风控模型比逻辑回归和决策树算法模型具有更好的分类效果和泛化能力。

Big data risk control refers to that financial institutions use machine learning methods to replace mortgage assets with behavioral big data and conduct credit risk assessment,so as to solve the problem of risk control for long tail customers.GBDT algorithm is an efficient integrated learning algorithm based on gradient boosting.In this paper,we build a big data risk control model based on GBDT algorithm,and conduct empirical research on the real personal credit data from the LendingClub platform.The results show that the risk control model based on GBDT algorithm has better classification effect and generalization ability than the logic regression model and decision tree model.