In this Repository i upload all task of internship .
This repository contains my submission for the Tata GenAI Powered Data Analytics Job Simulation on Forage, where I worked as a Data Analyst for a fictional financial services company, Geldium, to help predict and manage customer loan delinquency using data analysis, machine learning, and generative AI tools.
Delinquency — a customer failing to make timely payments — is a major risk for lenders. The goal of this project was to analyze customer financial and behavioral data, build a predictive model to flag customers likely to become delinquent, and translate those findings into actionable, responsible business recommendations for Geldium's Collections team.
-> Task 1 – Exploratory Data Analysis: Explored the customer dataset, checked data quality, and identified missing values and key patterns.
-> Task 2 – Predictive Model Plan: Built and trained a Logistic Regression model to predict delinquency, and documented the model logic, justification for the chosen algorithm, and an evaluation strategy.
-> Task 3 – Business Summary Report: Translated model insights into a business-facing summary with SMART recommendations and ethical/responsible AI considerations for the Collections team.
-> Task 4 – AI-Powered Collections Strategy: Designed a proposal for an agentic AI system to automate and scale collections outreach, with responsible AI guardrails and expected business impact.
Algorithm: Logistic Regression (scikit-learn) Target variable: Delinquent_Account (0 = Not Delinquent, 1 = Delinquent) Key features: Credit Score, Credit Utilization, Missed Payments, Debt-to-Income Ratio, Loan Balance, Employment Status, Account Tenure, and 6-month payment history Accuracy: 86% on a held-out test set (80/20 split)
-> Language :- Python
-> Libraries :- pandas, NumPy, scikit-learn
-> Tools :- Jupyter Notebook, GenAI tools (for model logic drafting and ideation)
├── intenship_submission.ipynb
├── Task_2_Predictive_Model_Plan.pdf
├── Task_3_Business_Summary_Report.pdf
├── Task_4_AI_Collections_Strategy.pptx / .pdf
└── README.md
Handling missing data and encoding categorical variables for machine learning models .
Building and evaluating a classification model for a real-world financial use case .
Communicating technical model results to a non-technical business audience .
Considering fairness, transparency, and ethical implications when deploying AI in financial decision-making .