Recurring Transaction Date Prediction

Context

For a Dutch bank, knowing when a customer’s recurring transactions (rent, subscriptions, salary) will hit is valuable for cash flow predictions, overdraft prevention, and personalized financial insights.

What I built

• Detected recurring transaction patterns from historical transaction data
• Predicted exact dates for future recurring transactions
• Achieved >80% exact accuracy where patterns were found

Tools & methods

• Python
• Pattern detection
• Time series