You are working with a financial services company that needs to analyze large volumes of transaction data for fraud detection. The data includes transaction details, user profiles, and timestamps. What partition strategy would you implement to optimize the performance of fraud detection algorithms?