Credit risk analysis and default prediction to support data-driven lending decisions.

Interview Coding Challenge Data Science Step 1 of the Data Scientist Interview process. Follow the instructions below to complete this portion of the interview. Please note, although we do not set a time limit for this challenge, we recommend completing it as soon as possible as we evaluate candidates on a first come, first serve basis... If you have any questions, please feel free to email support@TheZig.io. We will do our best to clarify any issues you come across. Prerequisites: A Text Editor - We recommend Visual Studio Code for the ClientSide code, its lightweight, powerful and Free! (https://code.visualstudio.com/) SQL Server Management Studio (https://docs.microsoft.com/en-us/sql/ssms/download-sql-server-management-studio-ssms?view=sql-server-2017) R - Software Environment for statitistal computing and graphics. You can download R at the mirrors listed here (https://cran.r-project.org/mirrors.html) Azure - Microsoft's Cloud Computing platform. You can create an account without a credit card by using the Azure Pass available at this link (https://azure.microsoft.com/en-us/offers/azure-pass/) Git - For source control and committing your final solution to a new private repo (https://git-scm.com/downloads) a. If you're not very familiar with git commands, here's a helpful cheatsheet (https://services.github.com/on-demand/downloads/github-git-cheat-sheet.pdf) 'R' Challenge For each numbered section below, write R code and comments to solve the problem or to show your rationale. For sections that ask you to give outputs, provide outputs in separate files and name them with the section number and the word output "Section 1 - Output". Create a private repo and submit your modified R script along with any supporting files. Load in the dataset from the accompanying file "account-defaults.csv" This dataset contains information about loan accounts that either went delinquent or stayed current on payments within the loan's first year. FirstYearDelinquency is the outcome variable, all others are predictors. The objective of modeling with this dataset is to be able to predict the probability that new accounts will become delinquent; it is primarily valuable to understand lower-risk accounts versus higher-risk accounts (and not just to predict 'yes' or 'no' for new accounts). FirstYearDelinquency - indicates whether the loan went delinquent within the first year of the loan's life (values of 1) AgeOldestIdentityRecord - number of months since the first record was reported by a national credit source AgeOldestAccount - number of months since the oldest account was opened AgeNewestAutoAccount - number of months since the most recent auto loan or lease account was opened TotalInquiries - total number of credit inquiries on record AvgAgeAutoAccounts - average number of months since auto loan or lease accounts were opened TotalAutoAccountsNeverDelinquent - total number of auto loan or lease accounts that were never delinquent WorstDelinquency - worst status of days-delinquent on an account in the first 12 months of an account's life; values of '400' indicate '400 or greater' HasInquiryTelecomm - indicates whether one or more telecommunications credit inquires are on record within the last 12 months (values of 1) Perform an exploratory data analysis on the accounts data In your analysis include summary statistics and visualizations of the distributions and relationships. Build one or more predictive model(s) on the accounts data using regression techniques Identify the strongest predictor variables and provide interpretations. Identify and explain issues with the model(s) such as collinearity, etc. Calculate predictions and show model performance on out-of-sample data. Summarize out-of-sample data in tiers from highest-risk to lowest-risk. Split up the dataset by the WorstDelinquency variable. For each subset, run a simple regression of FirstYearDelinquency ~ TotalInquiries. Extract the predictor's coefficient and p-value from each model. Store the in a list where the names of the list correspond to the values of WorstDelinquency. Load in the dataset from the accompanying file "vehicle-depreciation.csv". The dataset contains information about vehicles that our company purchases at auction, sells to customers, repossess from defaulted accounts, and finally re-sell at auction to recover some of our losses. Perform an analysis and/or build a predictive model that provides a method to estimate the depreciation of vehicle worth (from auction purchase to auction sale). Use whatever techniques you want to provide insight into the dataset and walk us through your results - this is your chance to show off your analytical and storytelling skills! CustomerGrade - the credit risk grade of the customer AuctionPurchaseDate - the date that the vehicle was purchased at auction AuctionPurchaseAmount - the dollar amount spent purchasing the vehicle at auction AuctionSaleDate - the date that the vehicle was sold at auction AuctionSaleAmount - the dollar amount received for selling the vehicle at auction VehicleType - the high-level class of the vehicle Year - the year of the vehicle Make - the make of the vehicle Model - the model of the vehicle Trim - the trim of the vehicle BodyType - the body style of the vehicle AuctionPurchaseOdometer - the odometer value of the vehicle at the time of purchase at the auction AutomaticTransmission - indicates (with value of 1) whether the vehicle has an automatic transmission DriveType - the drivetrain type of the vehicle

Risk analysis of credit card default using a data set provided by the Brazilian fintech Nubank.

The goal of this exercise is to analyze a dataset consisting of information from credit card holders and to comprehend which factors influence the Credit Card Balance of a cardholder and to predict the average Balance of a given individual. Such an exercise could be conducted as part of a customer analysis within a credit card company. The results of the analysis could determine which customers present a risk of credit default, or what the expected consumer behavior of prospective customers will be. In addition, combining the credit Balance data with information such as credit Limit can assist in calculating the credit utilization of a card, information which feeds into a cardholder's credit Rating. For this goal, a multivariable regression analysis will be undertaken. The exercise will begin with an exploratory data analysis of the dataset, followed by feature selection and regression analysis, including linear and logistic regression. Lastly, the regression model created will be employed to simulate a new dataset and predict the credit Balance of cardholders given their demographic information.

Credit card risk analysis. Whether the person is likely to be a defaulter or not..

The purpose of this project is to conduct quantitative analysis on credit card default risk by using 3 machine learning models with accessible customer data, instead of credit score or credit history, with the goal of assisting and speeding up the human decision-making process.

Credit Card Default Risk Prediction is an end-to-end machine learning project that predicts the likelihood of credit card default using a Logistic Regression model. The project includes detailed exploratory data analysis, feature engineering, class imbalance handling, and model evaluation, followed by deployment as an interactive web application.

Done as a part of NPCI 2022 Hackathon

No description available

My first financial risk project 🌅

This project presents a comprehensive descriptive and visual analysis of credit card customer behavior, with a focus on spending patterns, financial discipline, credit utilization, and default risk. The primary objective is to understand how different customer attributes—such as age, income, credit limits, card types, utilization rate, payment rati

This project is related with a loan and challenge is to predict approval status of loan (Approved/ Reject). Credit risk analysis and credit risk management is important to provide loans to businesses and individuals. Credit risk can occur for various reasons such as bank mortgages (or home loans), motor vehicle purchase finances, credit card purchases, installment purchases, and so on. Credit loans and finances have risk of being defaulted. To understand risk levels of credit users, we will (and credit providers) normally collect vast amount of information on borrowers. Some predictive analytic techniques can be used to analyze or to determine risk levels involved on credits, finances, and loans, i.e., default risk levels. We are trying to find default probability of Cumulative Accuracy Profile (CAP), the Receiver Operating Characteristic (ROC), and the Kolmogorov-Smirnov (K-S) statistic.

This project explores credit risk assessment using real-world credit card default data. The goal is to build a predictive model to determine the likelihood of default, using MySQL for data storage, Python for Analysis and PowerBI for visualisation. The project follows best practices to demonstrate an end-to-end financial DS workflow.

In banking sector credit risk plays an important role. The main operations of banks include loans, credit cards, deposits, mortgages, among others. From this Credit cards have become most common financial facilities provided by financial organizations in recent years. Though, financial institutions are confronting a rising credit defaulters’ ratio with an increasing number of credit or debit card customers. Data analysis will have the ability to provide solution to deal with this situation of credit defaulters. The research is focused on predicting an accuracy of defaulting payment of credit card users using data mining techniques. The data mining technique such as logistic regression and random forest are used to predict payment defaulter

This project focuses on analyzing customer credit data to identify key factors contributing to credit card default. Using the Application and Previous Application datasets, we aim to perform thorough Exploratory Data Analysis (EDA) and build predictive models to assist in risk assessment and decision-making for lending.

This repository contains data analysis and predictive modeling tools focused on credit card default payments in a banking context, using the UCI Machine Learning Repository dataset. The project leverages PySpark for large-scale data processing and model training, enabling scalable and efficient analysis of customer behavior and risk prediction.

credit card default risk analysis

No description available

No description available

Using decision tree classifier to classify credit card defaulter.

To develop your understanding of the domain, you are advised to independently research a little about risk analytics, understanding the types of variables and their significance should be enough.

Understanding the factors that influence credit card default risk is essential for financial institutions to mitigate losses and for individuals to manage their credit responsibly. This project focuses on analyzing credit card default risk using a dataset from UCI.

No description available

In Essence, the code cleans the data, prepares it for machine learning by scaling numerical features and splitting it into training/testing sets, and then trains and evaluates three different classification models (Logistic Regression, SVM, and Random Forest) to predict credit card default.

To analyze customer credit card behavior and build predictive models to identify default risk using demographic, financial, and behavioral features

No description available

No description available

This project analyzes the relationship between **age, credit limit, and default probability** using real-world banking data.

No description available

The risk analysis of credit card defaulters is a critical procedure in the banking sector to classify the card applicants. Banks perform credit score check to make decisions on applications and to set credit limit accordingly. With the increase in the amount of data and advances in data analytics, the approval process can now be automated for quicker processing of applications.