irene2024hub/Financial-Stress-Prediction-using-Machine-Learning

This model was developed using a mobile-money customer dataset provided as part of a Zindi Africa financial stress prediction competition. The dataset contains six months of customer transaction behaviour and demographic information, which are used to estimate the probability of liquidity stress within the following 30 days. # 💳 Financial Stress Prediction using Machine Learning ### *Early Detection of Customer Liquidity Risk from Mobile Money Behaviour* --- ## 📋 Executive Summary Across Africa and other emerging markets, mobile money has become the primary financial infrastructure for hundreds of millions of people — replacing traditional bank accounts for receiving income, paying bills, and managing daily life. This creates something traditional banking never had: a rich, continuous behavioural record of how a customer actually manages money, month after month. This project uses that behavioural record to answer a question that matters to every financial institution: **can we tell that a customer is heading toward financial difficulty *before* they miss a payment or default?** Traditional credit scoring is largely backward-looking — it flags risk only after damage is already visible (a missed payment, a defaulted loan). By contrast, **behavioural signals in transaction data — a declining balance, a shrinking financial network, growing withdrawal-to-deposit ratios — show up weeks before a formal default occurs.** This project builds a machine learning system that reads those early signals and outputs a calibrated, actionable probability of liquidity stress 30 days in advance, giving institutions a genuine window to intervene. Think of it as the difference between a smoke detector and a fire report. Traditional credit scoring tells an institution a fire already happened. This project is a smoke detector — it flags the early signs of financial difficulty while there's still time to act, support the customer, and prevent the loss altogether. That shift, from reactive to proactive, is the core business case for this entire project. --- ## 🎯 Business Problem Financial institutions — banks, digital lenders, and mobile money providers — lose value in two directions when they can't see financial stress coming: - **They act too late.** By the time a missed payment or loan defaul …