Text this: A hybrid approach to predicting and classifying dental impaction: integrating regularized regression and XG boost methods