When faced with a classification task involving complex and nonlinear decision boundaries in a data science project, which Spark ML algorithm would you choose to effectively model these intricate patterns?