This study presents a machine learning model to predict mortgage prepayment risks at the loan origination phase, leveraging variables such as loan-to-value ratios, credit scores and interest rates.