Exploring applicability of artificial intelligence and multivariate linear regression model for prediction of trihalomethanes in drinking water