A financial institution notices an unusual pattern in transactions where large amounts are transferred between new accounts shortly after opening. Which data-driven method would be most effective for detecting potential fraud in this scenario?

Explanation: Supervised machine learning (ML) is an effective approach to detect fraud by learning from labeled historical data of fraudulent and non-fraudulent transactions. In this case, ML models, such as decision trees, random forests, or gradient boosting machines, can analyze patterns in transaction history, account details, and timing to predict suspicious activities. These models can incorporate features like transaction frequency, location anomalies, and unusual amounts to make accurate predictions. By training on labeled data, supervised ML builds robust classifiers that reduce false positives and enhance fraud detection efficiency. This approach also adapts to new fraud techniques when retrained on updated datasets. Option A: Association rule mining identifies relationships between items (e.g., product purchases) but is not suited for fraud detection. Option B: Rule-based systems are static and may generate numerous false positives, failing to adapt to evolving fraud patterns. Option D: EDA identifies trends but is exploratory in nature and does not automate fraud detection. Option E: Dashboards visualize data but lack analytical capabilities for detecting fraud.