Small merchant credit evaluation method of fusion of expert model and machine learning model

Abstract

The invention provides a small merchant credit evaluation method of fusion of an expert model and a machine learning model. The method comprises six steps of data acquisition, data cleaning, arrangement, structuring, database inputting, middle layer characteristic extraction and quantification, expert model grading, machine learning model prediction, user final credit score outputting through fusion of two kinds of models and user rating result outputting. In the invention, because the machine learning model is added, a variable which seems to be unimportant can be used; a plurality of weak variables can become a strong variable after learning so that a condition that all the data become credits is possible; the two models are continuously fused and accurate and dynamic credit evaluation can be performed; and through a scoring result, a manual operation risk can be interpreted and avoided.

Description

technical field

[0001] The invention specifically relates to a small merchant credit evaluation method in which an expert model and a machine learning model are fused. Background technique

[0002] In the era of big data, various forms of Internet finance have combined modern information technology and applications represented by big data to build a new business model. If you explore the relationship between Internet finance and the popularity of big data, there are A very critical factor is that Internet finance has been unable to answer a core proposition - risk control. The core link of Internet finance is risk control, and the healthy growth of the industry also depends on it. Internet finance cannot simply bring the traditional financial service model online. Its core competitiveness is not the marketing ability to acquire customers but the big data risk control ability.

[0003] The traditional credit scoring model (hereinafter referred to as the expert model) is bas...

Images