Click to download ——Python Credit score card modeling （ The attached code ）
Extraction code : w6bt
Course overview ：
Python Credit score card modeling （ The attached code ）, Full video tutorial . In this course, students will learn to model the logistic regression score card as quickly as possible in one week , Save half a year to a year . The course offers practical projects , Students don't have to go to a large factory to experience real internship projects .
1. Project background
1.1 The basic concepts of credit risk and scorecard model
Credit risk refers to the risk of economic loss caused by the failure of the counterparties to fulfill their obligations in the contract , That is, the possibility that the trustee can not perform the obligation of repaying the principal and interest, and make the expected income of the credit giver deviate from the actual income , It's the main type of financial risk .
The score card in the lending scenario is a way to measure the risk probability in the form of score , It's also a breach of contract for a period of time in the future 、 Within the time limit 、 Prediction of the probability of loss of contact . Generally speaking , The higher the score , The less risk .
The credit risk measurement system includes two parts: subject rating model and debt rating . The main body rating model includes the following four aspects ：
Applicant rating model ： It is mainly applied to the subject rating of new users in related financing business , Applicable to individual and institutional financing entities . Located in the pre loan access link .
Behavior rating model ： It is mainly used for the management of stock customers in related financing business during the renewal period , For example, the customer may be overdue 、 Delay and so on , Only applicable to individual financing entities .
Collection rating model ： It is mainly used in the forecast management of whether stock customers need to be collected in related financing business , Only applicable to individual financing entities .
Fraud rating model ： It is mainly used to forecast and manage the possible fraud of new customers in related financing business , It is applicable to individuals and institutions . In the pre loan access link .
This project mainly aims at the scoring model of applicants .
1.2 Data sources
The data of this project comes from kaggle competition Give Me Some Credit.
2. Credit card scoring model development
The construction of the model mainly includes the following parts ： Data preparation and data preprocessing 、 Variable selection 、 model building 、 Model validation 、 Model to evaluate 、 Model deployment 、 The model monitors several parts . Let's talk about one by one .
2.1 Data preparation and data preprocessing
2.1.1 get data
Data acquisition includes stock customers, including data acquisition of stock customers and potential customers . Stock customers refer to the customers who have carried out relevant financing business in securities companies , Including individual clients and institutional clients ; Potential customers refer to the customers who plan to carry out relevant financing business in securities companies in the future , It mainly includes institutional clients , This is also a common way to solve the problem of less samples in the securities industry , These potential institutional clients include listed companies 、 The issuers of public bonds 、 New third board listed companies 、 Regional equity trading center listed companies 、 Non standard financing institutions, etc .