Lerato-Makutu/paysim-fraud-analysis

SQL analysis of mobile money transactions to detect fraud patterns using the PaySim dataset. # Mobile Money Transaction & Fraud Analysis (PaySim Dataset) ## Project Overview This project analyzes mobile money transactions to understand transaction behavior, identify fraud patterns, and provide recommendations for improving fraud detection systems. The analysis was conducted using SQL for data exploration and fraud detection, and Tableau to visualize transaction behavior and highlight key insights through an interactive dashboard. ## Objectives The goal of this analysis was to: - Understand overall transaction activity - Identify which transaction types are most common - Determine which transaction types move the most money - Detect fraud patterns within the system - Identify high-risk accounts and suspicious activity - Provide business recommendations to reduce fraud risk ## Dataset Dataset used: PaySim Mobile Money Transactions** The dataset simulates real mobile financial transactions and contains: - 199,999 transactions - Transaction types (payment, transfer, cash in, cash out, debit) - Transaction amounts - Sender and receiver accounts - Fraud indicators Key fields include: - `type` - `amount` - `nameOrig` - `nameDest` - `isFraud` ## Tools Used - SQL - MySQL - Tableau - GitHub ## Key Insights ### Transaction Activity - Total transactions: 199,999 - Total transaction value: R36,162,021,561.57 - Average transaction value: R180,811 - Largest transaction: R10,000,000 ### Transaction Behavior Most common transaction types: 1. Payment 2. Cash Out 3. Cash In 4. Transfer 5. Debit Although payments occur most frequently, transfers move the largest amount of money. ### Fraud Analysis Fraud was found in only two transaction types: - Transfer - Cash Out Fraud rates: - Transfer: 0.43% - Cash Out: 0.11% ## Example Analysis Output Below is a preview of the fraud analysis results from the SQL queries. Total financial loss due to fraud: R93,476,299.99 Average fraud transaction: R635,893 ### User Activity The analysis identified: - Top accou …